EP1849157B1 - Method of measuring annoyance caused by noise in an audio signal - Google Patents

Method of measuring annoyance caused by noise in an audio signal Download PDF

Info

Publication number
EP1849157B1
EP1849157B1 EP06709505A EP06709505A EP1849157B1 EP 1849157 B1 EP1849157 B1 EP 1849157B1 EP 06709505 A EP06709505 A EP 06709505A EP 06709505 A EP06709505 A EP 06709505A EP 1849157 B1 EP1849157 B1 EP 1849157B1
Authority
EP
European Patent Office
Prior art keywords
noise
signal
frame
calculating
frames
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Not-in-force
Application number
EP06709505A
Other languages
German (de)
French (fr)
Other versions
EP1849157A1 (en
Inventor
Nicolas Le Faucheur
Valérie GAUTIER-TURBIN
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Orange SA
Original Assignee
France Telecom SA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by France Telecom SA filed Critical France Telecom SA
Publication of EP1849157A1 publication Critical patent/EP1849157A1/en
Application granted granted Critical
Publication of EP1849157B1 publication Critical patent/EP1849157B1/en
Not-in-force legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L25/00Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
    • G10L25/48Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
    • G10L25/69Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for evaluating synthetic or decoded voice signals
    • GPHYSICS
    • G10MUSICAL INSTRUMENTS; ACOUSTICS
    • G10LSPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208Noise filtering

Definitions

  • the present invention is generally in the fields of speech signal processing and psychoacoustics. More specifically, the invention relates to a method and a device for objective evaluation of the annoyance due to noise in audio signals.
  • the invention makes it possible to note objectively the annoyance due to noise in an audio signal processed by a noise reduction function.
  • a noise reduction function also known as a noise canceling or denoising function
  • a noise canceling or denoising function is intended to reduce the background noise level in a voice communication, or having at least one component voice. It has a specific interest when one of the interlocutors of this communication is immersed in a noisy environment that greatly impairs the intelligibility of his voice.
  • the noise reduction algorithms are based on a continuous estimation of the background noise level from the incident signal and a speech activity detection to distinguish the noise periods only from those with the useful speech signal. A filtering of the incident speech signal corresponding to the noisy speech signal is then performed to reduce the noise contribution determined from the noise estimate.
  • the invention will be used to evaluate noise annoyance at the output of communication equipment implementing a noise reduction function, the invention also applies to noisy signals. not treated by such a function.
  • the case of use of the invention on any noisy audio signal is therefore a particular case of the more general case of use of the invention on an audio signal processed by a noise reduction function.
  • the present invention aims to overcome the disadvantages of the prior art by providing a method and an objective computing device of a score equivalent to the subjective score as indicated in the document "ITU-T Recommendation P.835", characterizing the annoyance due to the presence of noise in an audio signal.
  • the method according to the invention varies according to whether the invention is used on any noisy audio signal or on an audio signal processed by a noise reduction function, in particular in the parameters for calculating the objective score according to the invention.
  • two embodiments that can also be considered as two distinct processes are presented.
  • the second embodiment applying to any noisy audio signal, and more general than the first embodiment, is easily deduced therefrom.
  • the invention provides a method for calculating an objective noise noise score in an audio signal processed by a noise reduction function as defined in claim 1.
  • the step of calculating mean loudness densities and tone coefficients is followed by a step of calculating the averages S Y.
  • the coefficients of this linear combination have the advantage of being able to be recalculated if new subjective test data substantially modify the previously established correlation. This makes it possible to improve an objective model fed by the method according to the invention, of calculating the annoyance due to noise in an audio signal processed by a noise reduction function, by a simple reconfiguration of the parameters of the method.
  • the invention also relates to a method for calculating an objective score of noise annoyance in an audio signal as defined in claim 4.
  • This method has the same advantages as the previous method, but applies to any noisy audio signal.
  • the coefficients of this linear combination have the advantage of being able to be recalculated if new subjective test data substantially modify the previously established correlation. This makes it possible to improve an objective model fed by the method according to the invention, of calculating the annoyance due to the noise in an audio signal, by a simple reconfiguration of the parameters of the method.
  • the step of calculating loudness densities and tone coefficients is preceded by a voice activity detection step on the test signal, so as to determine if a current frame of the noisy signal, and of the processed signal in the case of the first method, is a "m_noise" frame containing only noise, or a "m_parole” frame containing speech, called a useful signal frame.
  • This voice activity detection step makes it possible to very simply separate the different types of frames of the noisy signal, and of the signal processed in the case of the first method, by the use of the test signal.
  • the invention also relates to a test equipment for evaluating an objective note of the annoyance due to noise in an audio signal, characterized in that it comprises means adapted to implement one or the other of the methods according to the invention.
  • the test equipment includes computer means and a computer program, said program comprising instructions adapted to implement one or the other of said methods, when it is executed by said computer means. .
  • the invention also relates to a computer program on an information carrier, comprising instructions adapted to the implementation of one or the other of the methods according to the invention, when the program is loaded and executed in a computer system.
  • Two embodiments of the method according to the invention are described hereinafter, the first being applied to an audio signal processed by a noise reduction function, and the second being applied to any noisy audio signal.
  • the principle of the method according to the invention is the same in these two embodiments, in particular the calculation method is exactly the same, but in the second embodiment the audio signal processed by a noise reduction function is taken equal at the noisy signal.
  • the second embodiment can indeed be considered as a special case of the first embodiment, with an inhibited noise reduction function.
  • the annoyance due to the presence of noise in an audio signal processed by a function of noise reduction is objectively evaluated in a test environment represented at figure 1 .
  • a test environment comprises a source of SSA audio signals delivering a test audio signal x (n) containing only the useful signal, that is to say devoid of noise, for example a speech signal, and a source noise SB delivering a predefined noise signal.
  • this predefined noise signal is added to the selected test signal x (n), as represented by the AD addition operator.
  • the audio signal resulting from this addition of noise to the test signal x (n) is denoted xb (n) and is designated by the expression "noisy signal”.
  • the noisy signal xb (n) then constitutes the input signal of a noise reduction module MRB implementing a noise reduction function outputting an audio signal y (n) designated by the expression "processed signal ".
  • the processed signal y (n) is therefore an audio signal containing useful signal and residual noise.
  • the processed signal y (n) is then delivered to an EQT test equipment implementing a method of objective evaluation of the annoyance due to the noise in the processed signal, according to the invention.
  • the method according to the invention is implemented in the EQT test equipment in the form of a computer program.
  • the EQT test equipment optionally comprises electronic hardware to implement the method according to the invention.
  • the test equipment EQT receives as input the test signal x (n) and the noisy signal xb (n).
  • the test equipment EQT outputs an evaluation result RES, which is an objective note NOB_MOS of the discomfort due to the presence of noise in the processed signal y (n).
  • the mode of calculation of this objective note NOB_MOS will be described below.
  • the aforementioned audio signals x (n), xb (n) and y (n) are signals sampled in a digital format, n designating a sample any. These signals are for example supposed to be sampled at the sampling frequency of 8 kHz (kilo Hertz).
  • the test signal x (n) is a speech signal devoid of noise.
  • the noisy signal xb (n) then represents the initial speech signal x (n) degraded by a noisy environment (background or ambient noise), and the signal y (n) represents the signal xb (n) after noise reduction.
  • the signal x (n) is generated in an anechoic chamber.
  • the signal x (n) can also be generated in a "quiet" room having an "average" reverberation time of less than 0.5 seconds.
  • the noisy signal xb (n) is obtained by adding a predetermined contribution of noise to the signal x (n).
  • the signal y (n) is obtained either at the output of a noise reduction algorithm implanted on a personal computer, or at the output of a noise reduction network equipment and in the latter case, the signal y (n) is taken at the level of a PCM encoder (pulse modulation and coding).
  • the method of calculating the objective note NOB_MOS of the annoyance due to the noise in the processed signal y (n) according to the invention is represented in the form of an algorithm comprising steps a1 to a7.
  • a first step a1 the signals x (n), xb (n) and y (n) are respectively divided into successive time windows called frames.
  • Each signal frame, denoted m contains a predetermined number of samples of the signal, step a1 therefore consists of a change in the rate of each of these signals.
  • the signals x (n), xb (n) and y (n) in frame rate respectively produce the signals x [m], xb [m], and y [m].
  • a speech activity detection is performed on the signal x [m] so as to determine whether each respective current frame of index m of the signals xb [m] and y [m], is a frame containing only noise, denoted "m_noise", or a frame containing speech, that is to say the useful signal, and noted “m_parole”. This determination is made by comparing the signals xb [m] and y [m] with the test signal x [m] devoid of noise.
  • DAV speech activity detection
  • Each silence frame of x [m] corresponds in fact to a noise frame for the signals xb [m] and y [m], while each speech frame of x [m] corresponds to a speech frame for the signals xb [m] and y [m].
  • a third step a3 loudness measurements are made on at least sets of y [m_noise], y [m_parole], xb [m_parole] frames from the previous step a2, and at least one set of frames of the signal y [m] at the output of step a1. For example, if 8 seconds of sampled test signal at 8 kHz is used, it will be possible to work on 250 fields y [m] of 256 samples of signal y (n). In addition, the tone coefficients of at least one set of y [m_noise] frames are measured.
  • the mean loudness densities are calculated S Xb ( m_parole ), S Y ( m_parole ), S Y ( m ), and S Y ( m_noise ) of respectively each of the frames xb [m_parole], y [m_parole], y [m] and y [m_noise] of sets of frames considered.
  • the tone coefficients ⁇ Y ( m_noise ) of each of the y [m_noise] frames of the considered set of y [m_noise] frames are calculated.
  • a fourth step a4 the respective averages are calculated S Xb _ word , S Y _ word , S Y , and S Y _ noise of medium loudness densities S Xb ( m _ speech ), S Y ( m _ speech ), S Y ( m ), and S Y ( m _ noise ) previously calculated on the respective sets considered frames xb [m_parole], y [m_parole], y [m] and y [m_noise].
  • the average ⁇ Y _ noise ⁇ Y tone coefficients (m _ noise) previously calculated for all considered frames y [m_noise] is also calculated.
  • the subjective test database is for example a database of scores obtained with groups of listeners according to "ITU-T Recommendation P.835", in which these notes are called background noise notes.
  • weighting coefficients by the use of a database of subjective tests is not essential for each step of calculating an objective score NOB. Indeed, these coefficients must be obtained prior to the first use of the process, and may be the same for all uses of the process. These coefficients are nevertheless likely to evolve when new subjective data come to feed the database of subjective tests used.
  • the annoyance due to the presence of noise in any noisy audio signal is evaluated objectively.
  • the same test environment is used as in the figure 1 , but by removing the MRB noise reduction module.
  • the audio signal source SSA delivers a test audio signal x (n) containing only the wanted signal, to which is added a predefined noise signal generated by the noise source SB, to obtain at the output of the addition operator AD a noisy signal xb (n).
  • test signal x (n) and the noisy signal xb (n) are then directly sent to the input of the test equipment EQT implementing a method of objective evaluation of the annoyance due to the noise in the noisy signal.
  • xb (n) according to the invention.
  • the signals x (n) and xb (n) are assumed to be sampled at the 8 kHz sampling rate.
  • the test equipment EQT outputs an evaluation result RES, which is an objective note NOB_MOS of the annoyance due to the presence of noise in the noisy signal xb (n).
  • the method for calculating the objective note NOB_MOS of the annoyance due to the noise in the noisy signal xb (n) according to the invention is represented in the form of an algorithm comprising steps b1 to b7. These steps are similar to steps a1 to a7 previously described in the first embodiment, and will therefore be a little less detailed. It should be noted that if we apply the calculation steps a3 to a7 with the signal y (n) equal to the signal xb (n) in the case of the first embodiment, we reach the second embodiment.
  • a first step b1 the signals x (n) and xb (n) are split into frames x [m] and xb [m] of time index m.
  • a third step b3 loudness measurements are made on at least sets of frames xb [m_noise] and xb [m_parole] from the previous step b2, and at least one set of frames of the signal xb [m] in exit from step b1.
  • the tone coefficients of at least one set of frames xb [m_noise] are measured.
  • the mean loudness densities are calculated S Xb ( m ).
  • S Xb ( m _ word ) and S Xb ( m_noise ) respectively of the frames xb [m], xb [m_parole] and xb [m_noise] of the sets of frames considered.
  • the tone coefficients ⁇ Xb ( m_noise ) of each frames xb [m_noise] of the considered set of frames xb [m_noise] are calculated.
  • a fourth step b4 the respective averages are calculated S Xb , S Xb _ word, and S Xb _ noise of medium loudness densities S Xb ( m ), S Xb ( m_parole ) and S Xb ( m_noise ) previously calculated on the respective sets considered frames xb [m], xb [m_parole] and xb [m_noise].
  • the mean ⁇ Xb _ noise of the tone coefficients ⁇ Xb ( m_noise ) previously calculated on the considered set of frames xb [m_noise] is also calculated.
  • obtaining the weighting coefficients by the use of a database of subjective tests is not indispensable at each step of calculating an objective score NOB.
  • the calculation according to the invention of the average loudness density S U (m) of a frame of any index m of a given audio signal u [m], comprises the steps c1 to c7 represented in FIG. figure 4 and explained below.
  • the calculation according to the invention of the tone coefficient ⁇ (m) of a frame of any index m of a given audio signal u [m] comprises the steps c1, c2, c3 and c8 represented in FIG. figure 4 and explained below.
  • the signal u [m] represents any of the signals x [m], xb [m], or y [m] defined above.
  • a windowing is applied to the frame of index m of the signal u [m], for example a windowing of Hanning, Hamming or equivalent type.
  • m for example a windowing of Hanning, Hamming or equivalent type.
  • a fast Fourier transform (FFT) is applied to the windowed frame u_w [m] and a corresponding frame U (m, f) in the frequency domain is accordingly obtained.
  • FFT fast Fourier transform
  • the power spectral density ⁇ U (m, f) of the frame U (m, f) is calculated. Such a calculation is known to those skilled in the art and will not, therefore, be detailed here.
  • step c8 is used to calculate the coefficient of tone, then at step c4 for calculating the average loudness of loudness S U (m), since for these two signals the two calculations are necessary.
  • step c4 for the other signals of steps a3 and b3, we go to step c4 for the calculation of the average loudness density S U (m).
  • the calculation of the tone coefficient is independent of the calculation of the mean loudness density S U (m), the two calculations can be carried out in parallel or one after the other.
  • step c4 a frequency conversion of the frequency axis at the Barks scale is applied to the power spectral density ⁇ U (m, f) obtained in the previous step, and a density is consequently obtained.
  • spectral power, B U (m, b) on the Barks scale, also called Bark spectrum.
  • B U (m, b) on the Barks scale, also called Bark spectrum.
  • step c5 the power spectral density on the Barks scale, B U (m, b), is subjected to a convolution with the spreading function commonly used in psychoacoustics, and a result is consequently obtained.
  • spectral density spread over the Barks scale denoted E U (m, b).
  • This step makes it possible to take into account the interaction of the adjacent critical bands.
  • the spread spectrum density E U (m, b) obtained previously is converted into loudness densities expressed in sones.
  • a calibration of the spectral density spread on the Barks scale, E U (m, b) is performed by the respective power scaling and loudness scaling factors commonly used in psychoacoustics.
  • the size obtained is then converted on the scale of the phones.
  • the conversion on the scale of the phones is carried out based on the isosonic curves (Fletcher curves) in accordance with the standard NF ISO 226 "Normal isosonic lines".
  • step c6 there is a number B of loudness density values, S U (m, b), of the frame of index m for the critical band b, where B is the number of critical bands considered in the Barks scale and the index b varying from 1 to B.
  • the average loudness of loudness S U (m) according to the invention of a frame of index m is therefore the average of the B loudness density values S U (m, b), of the frame of index m for a critical band b considered.
  • This calculation is done according to the principle defined by JD Johnston in his article " Transform coding of audio using the perceptual noise criteria of the journal IEEE Journal on selected areas in communications, vol. 6, no. 2, February 1988 ".
  • the tone coefficient ⁇ of a basic signal is a measure to show whether certain pure frequencies emerge from this signal. It is equivalent to a tonal density. Indeed, the more the tone coefficient ⁇ is close to 0, the more the signal is likened to noise. Conversely, the more the tone coefficient ⁇ is close to 1, the more the signal is component tonal majority. A tone coefficient ⁇ close to 1 attests to the presence of useful signal, or speech signal.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Signal Processing (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • Acoustics & Sound (AREA)
  • Multimedia (AREA)
  • Quality & Reliability (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
  • Signal Processing Not Specific To The Method Of Recording And Reproducing (AREA)
  • Noise Elimination (AREA)

Abstract

A method of computing an objective score (NOB) of annoyance caused by noise in an audio signal processed by a noise reduction function, said method including a preliminary step of obtaining a predefined test audio signal (x[m]) containing a wanted signal free of noise, a noisy signal (xb[m]) obtained by adding a predefined noise signal to said test signal (x[m]), and a processed signal (y[m]) obtained by applying the noise reduction function to said noisy signal (xb[m]), wherein said method further includes a step (a 3, a 4 ) of measuring the apparent loudness of frames of said noisy signal (xb[m]) and said processed signal (y[m]) and of measuring tonality coefficients of frames of said processed signal (y[m]).

Description

La présente invention se situe de manière générale dans les domaines du traitement du signal de parole et de la psychoacoustique. Plus précisément l'invention concerne un procédé et un dispositif d'évaluation objective de la gêne due au bruit dans des signaux audio.The present invention is generally in the fields of speech signal processing and psychoacoustics. More specifically, the invention relates to a method and a device for objective evaluation of the annoyance due to noise in audio signals.

L'invention permet notamment de noter objectivement la gêne due au bruit dans un signal audio traité par une fonction de réduction de bruit.The invention makes it possible to note objectively the annoyance due to noise in an audio signal processed by a noise reduction function.

Dans le domaine de la transmission de signaux audio, une fonction de réduction de bruit, aussi appelée fonction de suppression de bruit ou de débruitage, a pour objectif de réduire le niveau de bruit de fond dans une communication vocale, ou ayant au moins une composante vocale. Elle présente un intérêt spécifique lorsque l'un des interlocuteurs de cette communication est immergé dans un milieu bruité qui nuit fortement à l'intelligibilité de sa voix. Les algorithmes de réduction de bruit sont basés sur une estimation en continu du niveau du bruit de fond à partir du signal incident et d'une détection d'activité vocale permettant de distinguer les périodes de bruit seul de celles avec du signal de parole utile. Un filtrage du signal de parole incident, correspondant au signal de parole bruité, est ensuite effectué de façon à réduire la contribution du bruit déterminée à partir de l'estimée du bruit.In the field of audio signal transmission, a noise reduction function, also known as a noise canceling or denoising function, is intended to reduce the background noise level in a voice communication, or having at least one component voice. It has a specific interest when one of the interlocutors of this communication is immersed in a noisy environment that greatly impairs the intelligibility of his voice. The noise reduction algorithms are based on a continuous estimation of the background noise level from the incident signal and a speech activity detection to distinguish the noise periods only from those with the useful speech signal. A filtering of the incident speech signal corresponding to the noisy speech signal is then performed to reduce the noise contribution determined from the noise estimate.

La gêne due à la présence de bruit dans un signal audio traité par une telle fonction de réduction de bruit est évaluée aujourd'hui de manière subjective seulement en se basant sur l'exploitation de résultats de tests mis en oeuvre selon le document "Recommandation UIT-T P.835 (11/2003)". Cette évaluation est faite sur une échelle de type MOS, d'après l'anglais Mean Opinion Score, qui donne une note de un à cinq de la gêne due au bruit, appelée "background noise" dans ce même document.The annoyance due to the presence of noise in an audio signal processed by such a noise reduction function is subjectively assessed today only on the basis of the exploitation of test results implemented according to the document "Recommendation ITU -T P.835 (11/2003) ". This evaluation is done on a MOS scale, according to the English Mean Opinion Score, which gives a score of one to five of the annoyance due to noise, called "background noise" in this same document.

L'inconvénient majeur de cette technique d'évaluation est la nécessité de mettre en oeuvre des tests subjectifs, cette mise en oeuvre étant très lourde et très coûteuse. En effet chaque contexte particulier, c'est-à-dire un type de signal incident associé à un type de bruit et une fonction de réduction de bruit, nécessite de mettre un panel de personnes en situation d'écoute réelle d'échantillons de parole afin de leur demander de noter la gêne due au bruit selon une échelle de type MOS.The major disadvantage of this evaluation technique is the need to implement subjective tests, this implementation being very heavy and very expensive. Indeed, each particular context, that is to say a type of incident signal associated with a type of noise and a noise reduction function, requires putting a panel of people in situation of actual listening of speech samples. to ask them to note the annoyance due to noise on a MOS scale.

C'est pourquoi le développement de méthodes objectives alternatives pouvant compléter ou suppléer les méthodes subjectives est un sujet de grand intérêt. L'illustration la plus frappante de ce phénomène est le modèle de qualité d'écoute, en constante évolution, contenu dans le document " Recommandation UIT-T P.862 (02/2001 )". Néanmoins ce modèle ne s'applique pas à l'évaluation de la gêne due au bruit. L'invention concerne en effet des signaux de parole dans lesquels la gêne due au bruit peut être importante, ceci avant ou après traitement de ces signaux par une éventuelle fonction de réduction de bruit.That is why the development of alternative objective methods that can complement or supplement subjective methods is a subject of great interest. The most striking illustration of this phenomenon is the model of listening quality, in constant evolution, contained in the document " ITU-T Recommendation P.862 (02/2001) However, this model does not apply to the evaluation of noise annoyance.The invention relates in fact to speech signals in which annoyance due to noise can be important, this before or after treatment of these signals by a possible noise reduction function.

Il est de plus à noter que bien qu'en général l'invention sera utilisée pour évaluer la gêne due au bruit en sortie d'un équipement de communication implémentant une fonction de réduction de bruit, l'invention s'applique aussi aux signaux bruités non traités par une telle fonction. Le cas d'utilisation de l'invention sur un signal audio bruité quelconque est donc un cas particulier du cas plus général d'utilisation de l'invention sur un signal audio traité par une fonction de réduction de bruit.It should also be noted that, although in general the invention will be used to evaluate noise annoyance at the output of communication equipment implementing a noise reduction function, the invention also applies to noisy signals. not treated by such a function. The case of use of the invention on any noisy audio signal is therefore a particular case of the more general case of use of the invention on an audio signal processed by a noise reduction function.

La présente invention a pour but de résoudre les inconvénients de la technique antérieure en fournissant un procédé et un dispositif de calcul objectif d'une note équivalente à la note subjective telle qu'indiquée dans le document "Recommandation UIT-T P.835", caractérisant la gêne due à la présence de bruit dans un signal audio. Le procédé selon l'invention varie suivant que l'invention est utilisé sur un signal audio bruité quelconque ou sur un signal audio traité par une fonction de réduction de bruit, notamment dans les paramètres de calcul de la note objective selon l'invention. Afin de bien décrire ces deux cas d'utilisation, deux modes de réalisation pouvant aussi être considérés comme deux procédés distincts sont présentés. Cependant le second mode de réalisation, s'appliquant à un signal audio bruité quelconque, et plus général que le premier mode de réalisation, se déduit aisément de celui-ci.The present invention aims to overcome the disadvantages of the prior art by providing a method and an objective computing device of a score equivalent to the subjective score as indicated in the document "ITU-T Recommendation P.835", characterizing the annoyance due to the presence of noise in an audio signal. The method according to the invention varies according to whether the invention is used on any noisy audio signal or on an audio signal processed by a noise reduction function, in particular in the parameters for calculating the objective score according to the invention. In order to describe these two use cases, two embodiments that can also be considered as two distinct processes are presented. However, the second embodiment, applying to any noisy audio signal, and more general than the first embodiment, is easily deduced therefrom.

A cette fin, l'invention propose un procédé de calcul d'une note objective de la gêne due au bruit dans un signal audio traité par une fonction de réduction de bruit, comme défini dans la revendication 1.To this end, the invention provides a method for calculating an objective noise noise score in an audio signal processed by a noise reduction function as defined in claim 1.

Ce procédé a l'avantage d'une mise en oeuvre simple, immédiate et rapide contrairement aux tests subjectifs. On rappellera ici que l'expression "sonie psychoacoustique" peut être définie comme le caractère de la sensation auditive lié à la pression acoustique et à la structure du son. En d'autres termes, il s'agit de la force sonore d'un son ou d'un bruit en tant que sensation auditive (cf. Office de la langue française, 1988). La sonie est représentée par une échelle de sonie psychoacoustique (en sones). D'autre part, la densité de sonie, encore désignée par "intensité subjective", est une mesure particulière de la sonie.This method has the advantage of simple, immediate and rapid implementation, contrary to subjective tests. It will be recalled here that the expression "psychoacoustic sony" can be defined as the character of the auditory sensation related to the sound pressure and the structure of the sound. In other words, it is the sound force of a sound or a sound as an auditory sensation (see Office de la langue française, 1988). The loudness is represented by a psychoacoustic loudness scale (in sones). On the other hand, loudness, still referred to as "subjective intensity," is a particular measure of loudness.

Selon une caractéristique préférée, ce procédé selon l'invention comprend les étapes de :

  • Calcul de densités de sonie moyenne S Y (m) de trames du signal traité, de densités de sonie moyenne respectives S Xb (m_parole) et S Y (m_parole) de trames de signal utile "m_parole" respectivement du signal bruité et du signal traité, de densités de sonie moyenne S Y (m_bruit) de trames de bruit "m_bruit" du signal traité, et de coefficients de tonalité α Y (m_bruit) de trames de bruit "m_bruit" du signal traité,
  • Calcul d'une note objective de la gêne due au bruit dans le signal traité, à partir desdites densités de sonie moyenne et desdits coefficients de tonalité calculés, et de coefficients de pondération prédéfinis.
According to a preferred feature, this method according to the invention comprises the steps of:
  • Calculation of average loudness S Y ( m ) of frames of the processed signal, respective mean loudness densities S Xb ( m_parole ) and S Y ( m_parole ) of useful signal frames "m_parole" respectively of the noisy signal and processed signal, average loudness of loudness S Y (m_noise) of noise frames "m_noise" of the processed signal, and tone coefficients α Y (m _ noise) of noise frames "m_noise" of the processed signal,
  • Calculating an objective note of noise annoyance in the processed signal from said average loudness densities and said calculated tone coefficients, and predefined weighting coefficients.

Selon une caractéristique préférée, l'étape de calcul de densités de sonie moyenne et de coefficients de tonalité est suivie d'une étape de calcul des moyennes S Y . S Xb _ parole , S Y _ parole, S Y _ bruit et αγ _ bruit desdites densités de sonie moyenne et desdits coefficients de tonalité sur l'ensemble des trames concernées des signaux correspondants, et la note objective de la gêne due au bruit est calculée selon l'équation suivante: NOB = i = 1 5 ω i facteur i + ω 6 ,

Figure imgb0001

facteur 1 = S Y _bruit S Y ,
Figure imgb0002

facteur 2 = S Y _bruit S Y _parole ,
Figure imgb0003

facteur(3)= Ecart_type ( S Xb (m_parole) - S Y(m_parole)), l'opérateur "Ecart_type (v(m))" désignant l'écart-type de la variable v sur l'ensemble des trames d'indice m,
facteur(4)= α Y _ bruit ,
facteur(5)= Ecart_type (α Y (m_bruit)),
et les coefficients ω1 à ω6 sont déterminés de manière à obtenir une corrélation maximale entre les données subjectives issues d'une base de données de tests subjectifs et les notes objectives calculées par ledit procédé pour les signaux de tests, bruités et traités correspondants utilisés lors desdits tests subjectifs.According to a preferred characteristic, the step of calculating mean loudness densities and tone coefficients is followed by a step of calculating the averages S Y. S Xb _ word , S Y _ word , S Y _ noise and α γ _ noise of said average loudness densities and of said tone coefficients on all the relevant frames of the corresponding signals, and the objective note of the annoyance due to the noise is calculated according to the following equation: NOB = Σ i = 1 5 ω i postman i + ω 6 ,
Figure imgb0001

or postman 1 = S ~ Y _noise S ~ Y ,
Figure imgb0002

postman 2 = S ~ Y _noise S ~ Y _parole ,
Figure imgb0003

factor (3) = Standard deviation ( S Xb ( m_parole ) - S Y (m_parole )), the operator "Ecart_type (v (m))" designating the standard deviation of the variable v on the set of frames of index m,
factor (4) = α Y _ noise ,
factor (5) = Standard deviation (α Y ( m_noise ) ) ,
and the coefficients ω 1 to ω 6 are determined so as to obtain a maximum correlation between the subjective data from a subjective test database and the objective scores calculated by said method for the test signals, noises and corresponding processes used. during said subjective tests.

Les coefficients de cette combinaison linéaire ont l'avantage de pouvoir être recalculés si de nouvelles données de tests subjectifs modifient de manière sensible la corrélation précédemment établie. Ceci permet d'améliorer un modèle objectif alimenté par le procédé selon l'invention, de calcul de la gêne due au bruit dans un signal audio traité par une fonction de réduction de bruit, par une simple reconfiguration des paramètres du procédé.The coefficients of this linear combination have the advantage of being able to be recalculated if new subjective test data substantially modify the previously established correlation. This makes it possible to improve an objective model fed by the method according to the invention, of calculating the annoyance due to noise in an audio signal processed by a noise reduction function, by a simple reconfiguration of the parameters of the method.

L'invention concerne aussi un procédé de calcul d'une note objective de la gêne due au bruit dans un signal audio, comme défini dans la revendication 4.The invention also relates to a method for calculating an objective score of noise annoyance in an audio signal as defined in claim 4.

Ce procédé a les mêmes avantages que le procédé précédent, mais s'applique à un signal audio bruité quelconque.This method has the same advantages as the previous method, but applies to any noisy audio signal.

Selon une caractéristique préférée, ce procédé selon l'invention comporte les étapes de:

  • Calcul de densités de sonie moyenne S Xb (m) de trames du signal bruité, de densités de sonie moyenne S Xb (m_parole) de trames de signal utile "m_parole" du signal bruité, de densités de sonie moyenne S Xb (m_bruit) de trames de bruit "m_bruit" du signal bruité, et de coefficients de tonalité α Xb (m_bruit) de trames de bruit "m_bruit" du signal bruité,
  • Calcul d'une note objective de la gêne due au bruit dans le signal bruité, à partir desdites densités de sonie moyennes et desdits coefficients de tonalité calculés, et de coefficients de pondération prédéfinis.
According to a preferred characteristic, this method according to the invention comprises the steps of:
  • Calculation of average loudness S Xb ( m ) of noisy signal frames, average loudness S Xb ( m_parole ) of useful signal frames "m_parole" of the noisy signal, of medium loudness densities S Xb ( m_noise ) of noise frames "m_noise" of the noisy signal, and tone coefficients α Xb ( m_noise ) of noise frames "m_noise" of the noisy signal,
  • Calculating an objective score of noise annoyance in the noisy signal from said average loudness densities and said calculated tone coefficients, and predefined weighting coefficients.

Selon une caractéristique préférée, l'étape de calcul de densités de sonie moyenne et de coefficients de tonalité est suivie d'une étape de calcul des moyennes S Xb , S Xb _ parole , S Xb _ bruit et α Xb _ bruit desdites densités de sonie moyenne et desdits coefficients de tonalité sur l'ensemble des trames concernées des signaux correspondants, et en ce que ladite note objective de la gêne due au bruit est calculée selon l'équation suivante: NOB = i = 1 4 ω i facteur i + ω 5 ,

Figure imgb0004


facteur 1 = S Xb _bruit S Xb ,
Figure imgb0005

facteur 2 = S Xb _bruit S Xb _parole ,
Figure imgb0006

facteur(3)=αXb_bruit,
facteur(4)= Ecart_type(α Xb (m_bruit)), l'opérateur "Ecart_type (v(m))" désignant l'écart-type de la variable v sur l'ensemble des trames d'indice m,
et les coefficients ω1 à ω5 sont déterminés de manière à obtenir une corrélation maximale entre les données subjectives issues d'une base de données de tests subjectifs et les notes objectives calculées par ledit procédé pour les signaux de tests et les signaux bruités correspondants utilisés lors desdits tests subjectifs.According to a preferred characteristic, the step of calculating mean loudness densities and tone coefficients is followed by a calculation step averages S Xb , S Xb _ word , S Xb _ noise and α Xb _ noise of said average loudness densities and said tone coefficients on all the relevant frames of the corresponding signals, and in that said objective note of the annoyance due to the noise is calculated according to the following equation: NOB = Σ i = 1 4 ω i postman i + ω 5 ,
Figure imgb0004

or
postman 1 = S ~ Xb _noise S ~ Xb ,
Figure imgb0005

postman 2 = S ~ Xb _noise S ~ Xb _parole ,
Figure imgb0006

factor (3) = α Xb_noise ,
factor (4) = STDEV (α Xb (m _ noise)), the operator "STDEV (v (m))" designating the standard deviation of the variable v over all the frames of index m,
and the coefficients ω 1 to ω 5 are determined to obtain maximum correlation between the subjective data from a subjective test database and the objective scores calculated by said method for the test signals and the corresponding noisy signals used. during said subjective tests.

Comme pour le procédé précédent, les coefficients de cette combinaison linéaire ont l'avantage de pouvoir être recalculés si de nouvelles données de tests subjectifs modifient de manière sensible la corrélation précédemment établie. Ceci permet d'améliorer un modèle objectif alimenté par le procédé selon l'invention, de calcul de la gêne due au bruit dans un signal audio, par une simple reconfiguration des paramètres du procédé.As for the previous method, the coefficients of this linear combination have the advantage of being able to be recalculated if new subjective test data substantially modify the previously established correlation. This makes it possible to improve an objective model fed by the method according to the invention, of calculating the annoyance due to the noise in an audio signal, by a simple reconfiguration of the parameters of the method.

Selon une caractéristique préférée de ces deux procédés selon l'invention, l'étape de calcul de densités de sonie et de coefficients de tonalité est précédée d'une étape de détection d'activité vocale sur le signal de test, de manière à déterminer si une trame courante du signal bruité, et du signal traité dans le cas du premier procédé, est une trame "m_bruit" contenant seulement du bruit, ou une trame "m_parole" contenant de la parole, dite trame de signal utile.According to a preferred characteristic of these two methods according to the invention, the step of calculating loudness densities and tone coefficients is preceded by a voice activity detection step on the test signal, so as to determine if a current frame of the noisy signal, and of the processed signal in the case of the first method, is a "m_noise" frame containing only noise, or a "m_parole" frame containing speech, called a useful signal frame.

Cette étape de détection d'activité vocale permet de séparer très simplement les différents types de trames du signal bruité, et du signal traité dans le cas du premier procédé, par l'utilisation du signal de test.This voice activity detection step makes it possible to very simply separate the different types of frames of the noisy signal, and of the signal processed in the case of the first method, by the use of the test signal.

Selon une caractéristique préférée de ces deux procédés selon l'invention, l'étape de calcul de la note objective est suivie d'une étape de calcul d'une note objective sur l'échelle MOS de la gêne due au bruit, calculée selon l'équation suivante: NOB_MOS = i = 1 4 λ i NOB i - 1 ,

Figure imgb0007

dans laquelle les coefficients λ1 à λ4 sont déterminés de manière à ce que ladite nouvelle note objective obtenue caractérise la gêne due au bruit sur l'échelle MOS.According to a preferred characteristic of these two methods according to the invention, the step of calculating the objective score is followed by a step of calculating an objective score on the MOS scale of the annoyance due to the noise, calculated according to the following equation: NOB_MOS = Σ i = 1 4 λ i NOB i - 1 ,
Figure imgb0007

wherein the coefficients λ 1 to λ 4 are determined in such a way that said new objective score obtained characterizes the annoyance due to the noise on the MOS scale.

Le fait d'utiliser une fonction polynomiale d'ordre 3 permet d'obtenir une note objective sur l'échelle MOS très proche de la note subjective MOS que donnerait un groupe d'auditeurs dans le cadre d'un test subjectif conforme à la "Recommandation UIT-T P.835".The fact of using a polynomial function of order 3 makes it possible to obtain an objective score on the MOS scale very close to the subjective MOS score that would be given by a group of listeners in the context of a subjective test conforming to the " ITU-T Recommendation P.835 ".

Selon une caractéristique préférée de ces deux procédés selon l'invention, l'étape de calcul de densités de sonie et de coefficients de tonalité, le calcul de la densité de sonie moyenne S U(m) d'une trame d'indice m quelconque d'un signal audio donné u, comprend les étapes suivantes :

  • fenêtrage, par exemple de type Hanning, de la trame d'indice m et obtention d'une trame fenêtrée u_w[m],
  • application d'une transformée de Fourier rapide à la trame fenêtrée u_w[m] et obtention d'une trame correspondante U(m,f) dans le domaine fréquentiel,
  • calcul de la densité spectrale de puissance γU(m, f) de la trame U(m,f),
  • application à la densité spectrale de puissance γU(m,f) d'une conversion de l'axe des fréquences à l'échelle des Barks et obtention d'une densité spectrale de puissance BU(m, b) sur l'échelle des Barks,
  • convolution de la densité spectrale de puissance sur l'échelle des Barks, BU(m, b), avec la fonction d'étalement couramment utilisée en psychoacoustique et obtention d'une densité spectrale étalée sur l'échelle des Barks, EU(m,b),
  • calibration de la densité spectrale étalée sur l'échelle des Barks, EU(m,b), par les facteurs respectifs d'échelonnement en puissance et d'échelonnement en sonie couramment utilisés en psychoacoustique, conversion de la grandeur ainsi obtenue sur l'échelle des phones puis conversion sur l'échelle des sones de la grandeur précédemment convertie en phones, et obtention en conséquence d'un nombre B de valeurs de densité de sonie, SU(m, b), de la trame d'indice m pour la bande critique b, B étant le nombre de bandes critiques considérées dans l'échelle des Barks et l'indice b variant de 1 à B ,
  • calcul de la densité de sonie moyenne S U (m) de la trame d'indice m à partir desdites B valeurs de densités de sonie SU(m, b), selon l'équation suivante : S U m = 1 B b = 1 B S U m b
    Figure imgb0008
According to a preferred feature of these two methods according to the invention, the step of calculating loudness densities and tone coefficients, calculating the average loudness density S U (m) of a frame of any index m of a given audio signal u, comprises the following steps:
  • windowing, for example of the Hanning type, of the frame of index m and obtaining a windowed frame u_w [m],
  • applying a fast Fourier transform to the windowed frame u_w [m] and obtaining a corresponding frame U (m, f) in the frequency domain,
  • calculating the power spectral density γ U (m, f) of the frame U (m, f),
  • application to the power spectral density γ U (m, f) of a conversion of the frequency axis to the Barks scale and obtaining a spectral power density B U (m, b) on the scale Barks,
  • convolution of the spectral power density on the Barks scale, B U (m, b), with the spreading function commonly used in psychoacoustics and obtaining a spectral density spread on the Barks scale, E U ( m, b),
  • calibration of the spectral density spread on the Barks scale, E U (m, b), by the respective factors of power scaling and loudness scaling commonly used in psychoacoustics, conversion of the magnitude thus obtained on the scale of the phones then conversion on the sonic scale of the quantity previously converted to phones, and consequently obtaining a number B of loudness density values, S U (m, b), of the frame of index m for the critical band b, where B is the number of critical bands considered in the Barks scale and the index b ranging from 1 to B,
  • calculating the average loudness of loudness S U (m) of the frame of index m from said B loudness density values S U (m, b), according to the following equation: S ~ U m = 1 B Σ b = 1 B S U m b
    Figure imgb0008

Selon une caractéristique préférée de ces deux procédés selon l'invention, dans l'étape de calcul de densités de sonie et de coefficients de tonalité, le calcul du coefficient de tonalité α(m) d'une trame d'indice m quelconque d'un signal audio donné u, comprend les étapes suivantes :

  • fenêtrage, par exemple de type Hanning, de la trame d'indice m et obtention d'une trame fenêtrée u_w[m],
  • application d'une transformée de Fourier rapide à la trame fenêtrée u_w[m] et obtention d'une trame correspondante U(m,f) dans le domaine fréquentiel,
  • calcul de la densité spectrale de puissance γU(m, f) de la trame U(m,f),
  • calcul du coefficient de tonalité α(m) selon l'équation suivante: α m = 10 * log 10 f = 0 N - 1 γ U m f 1 / N 1 N f = 0 N - 1 γ U m f - 60 ,
    Figure imgb0009
où * symbolise l'opérateur de multiplication dans l'espace des nombres réels, f représente l'indice fréquentiel de la densité spectrale de puissance, et N désigne la taille de la transformée de Fourier rapide.According to a preferred characteristic of these two methods according to the invention, in the step of calculating loudness densities and tone coefficients, the calculation of the tone coefficient α (m) of a frame of any index m of a given audio signal u, comprises the following steps:
  • windowing, for example of the Hanning type, of the frame of index m and obtaining a windowed frame u_w [m],
  • applying a fast Fourier transform to the windowed frame u_w [m] and obtaining a corresponding frame U (m, f) in the frequency domain,
  • calculating the power spectral density γ U (m, f) of the frame U (m, f),
  • calculating the tone coefficient α (m) according to the following equation: α m = 10 * log 10 Π f = 0 NOT - 1 γ U m f 1 / NOT 1 NOT Σ f = 0 NOT - 1 γ U m f - 60 ,
    Figure imgb0009
where * symbolizes the multiplication operator in the real number space, f represents the frequency index of the power spectral density, and N denotes the size of the fast Fourier transform.

L'invention concerne également un équipement de test destiné à évaluer une note objective de la gêne due au bruit dans un signal audio, caractérisé en ce qu'il comporte des moyens adaptés à mettre en oeuvre l'un ou l'autre des procédés selon l'invention.The invention also relates to a test equipment for evaluating an objective note of the annoyance due to noise in an audio signal, characterized in that it comprises means adapted to implement one or the other of the methods according to the invention.

Selon une caractéristique préférée, l'équipement de test inclut des moyens informatiques et un programme d'ordinateur, ledit programme comportant des instructions adaptées à mettre en oeuvre l'un ou l'autre desdits procédés, lorsqu'il est exécuté par lesdits moyens informatiques.According to a preferred characteristic, the test equipment includes computer means and a computer program, said program comprising instructions adapted to implement one or the other of said methods, when it is executed by said computer means. .

L'invention concerne encore un programme d'ordinateur sur un support d'informations, comportant des instructions adaptées à la mise en oeuvre de l'un ou l'autre des procédés selon l'invention, lorsque le programme est chargé et exécuté dans un système informatique.The invention also relates to a computer program on an information carrier, comprising instructions adapted to the implementation of one or the other of the methods according to the invention, when the program is loaded and executed in a computer system.

Les avantages de cet équipement de test ou de ce programme d'ordinateur sont identiques à ceux mentionnés plus haut en relation avec les procédés de l'invention.The advantages of this test equipment or computer program are identical to those mentioned above in connection with the methods of the invention.

D'autres caractéristiques et avantages apparaîtront à la lecture de modes de réalisation préférés décrits en référence aux figures dans lesquelles:

  • la figure 1 représente un environnement de test destiné à calculer une note objective de la gêne due au bruit dans un signal audio traité par une fonction de réduction de bruit, selon un premier mode de réalisation de l'invention,
  • la figure 2 est un organigramme illustrant un procédé de calcul d'une note objective de la gêne due au bruit dans un signal audio traité par une fonction de réduction de bruit selon un premier mode de réalisation du procédé selon l'invention,
  • la figure 3 est un organigramme illustrant un procédé de calcul d'une note objective de la gêne due au bruit dans un signal audio selon un second mode de réalisation du procédé selon l'invention,
  • la figure 4 est un organigramme illustrant le mode de calcul de la densité de sonie moyenne et du coefficient de tonalité d'une trame de signal audio selon l'invention.
Other features and advantages will appear on reading preferred embodiments described with reference to the figures in which:
  • the figure 1 represents a test environment for calculating an objective score of noise annoyance in an audio signal processed by a noise reduction function, according to a first embodiment of the invention,
  • the figure 2 is a flowchart illustrating a method for calculating an objective note of the annoyance due to noise in an audio signal processed by a noise reduction function according to a first embodiment of the method according to the invention,
  • the figure 3 is a flowchart illustrating a method for calculating an objective note of the annoyance due to noise in an audio signal according to a second embodiment of the method according to the invention,
  • the figure 4 is a flowchart illustrating the method of calculating the average loudness density and the tone coefficient of an audio signal frame according to the invention.

Deux modes de réalisation du procédé selon l'invention sont décrits dans la suite, le premier étant appliqué à un signal audio traité par une fonction de réduction de bruit, et le second étant appliqué à un signal audio bruité quelconque. Le principe du procédé selon l'invention est le même dans ces deux modes de réalisation, en particulier le procédé de calcul est exactement le même, mais dans le second mode de réalisation le signal audio traité par une fonction de réduction de bruit est pris égal au signal bruité. Le second mode de réalisation peut en effet être considéré comme un cas particulier du premier mode de réalisation, avec une fonction de réduction de bruit inhibée.Two embodiments of the method according to the invention are described hereinafter, the first being applied to an audio signal processed by a noise reduction function, and the second being applied to any noisy audio signal. The principle of the method according to the invention is the same in these two embodiments, in particular the calculation method is exactly the same, but in the second embodiment the audio signal processed by a noise reduction function is taken equal at the noisy signal. The second embodiment can indeed be considered as a special case of the first embodiment, with an inhibited noise reduction function.

Selon le premier mode de réalisation du procédé l'invention, la gêne due à la présence de bruit dans un signal audio traité par une fonction de réduction de bruit est évaluée de manière objective dans un environnement de test représenté à la figure 1 . Un tel environnement de test comprend une source de signaux audio SSA délivrant un signal audio de test x(n) ne contenant que du signal utile, c'est-à-dire dépourvu de bruit, par exemple un signal de parole, et une source de bruit SB délivrant un signal de bruit prédéfini.According to the first embodiment of the method of the invention, the annoyance due to the presence of noise in an audio signal processed by a function of noise reduction is objectively evaluated in a test environment represented at figure 1 . Such a test environment comprises a source of SSA audio signals delivering a test audio signal x (n) containing only the useful signal, that is to say devoid of noise, for example a speech signal, and a source noise SB delivering a predefined noise signal.

Aux fins de test, ce signal de bruit prédéfini est ajouté au signal de test x(n) choisi, comme représenté par l'opérateur d'addition AD. Le signal audio résultant de cette addition de bruit au signal de test x(n) est noté xb(n) et est désigné par l'expression "signal bruité".For testing purposes, this predefined noise signal is added to the selected test signal x (n), as represented by the AD addition operator. The audio signal resulting from this addition of noise to the test signal x (n) is denoted xb (n) and is designated by the expression "noisy signal".

Le signal bruité xb(n) constitue alors le signal d'entrée d'un module MRB de réduction de bruit mettant en oeuvre une fonction de réduction de bruit délivrant en sortie un signal audio y(n) désigné par l'expression "signal traité". Le signal traité y(n) est donc un signal audio contenant du signal utile et un bruit résiduel.The noisy signal xb (n) then constitutes the input signal of a noise reduction module MRB implementing a noise reduction function outputting an audio signal y (n) designated by the expression "processed signal ". The processed signal y (n) is therefore an audio signal containing useful signal and residual noise.

Le signal traité y(n) est ensuite délivré à un équipement de test EQT mettant en oeuvre un procédé d'évaluation objective de la gêne due au bruit dans le signal traité, selon l'invention. Typiquement le procédé selon l'invention est implémenté dans l'équipement de test EQT sous la forme d'un programme d'ordinateur. En plus ou en remplacement de moyens logiciels, l'équipement de test EQT comporte éventuellement des moyens matériels électroniques pour implémenter le procédé selon l'invention. Outre le signal y(n), l'équipement de test EQT reçoit en entrée le signal de test x(n) et le signal bruité xb(n).The processed signal y (n) is then delivered to an EQT test equipment implementing a method of objective evaluation of the annoyance due to the noise in the processed signal, according to the invention. Typically the method according to the invention is implemented in the EQT test equipment in the form of a computer program. In addition to or in replacement of software means, the EQT test equipment optionally comprises electronic hardware to implement the method according to the invention. In addition to the signal y (n), the test equipment EQT receives as input the test signal x (n) and the noisy signal xb (n).

L'équipement de test EQT délivre en sortie un résultat d'évaluation RES, qui est une note objective NOB_MOS de la gêne due à la présence de bruit dans le signal traité y(n). Le mode de calcul de cette note objective NOB_MOS sera décrit plus bas.The test equipment EQT outputs an evaluation result RES, which is an objective note NOB_MOS of the discomfort due to the presence of noise in the processed signal y (n). The mode of calculation of this objective note NOB_MOS will be described below.

Les signaux audio précités x(n), xb(n) et y(n) sont des signaux échantillonnés dans un format numérique, n désignant un échantillon quelconque. Ces signaux sont par exemple supposés échantillonnés à la fréquence d'échantillonnage de 8 kHz (kilo Hertz).The aforementioned audio signals x (n), xb (n) and y (n) are signals sampled in a digital format, n designating a sample any. These signals are for example supposed to be sampled at the sampling frequency of 8 kHz (kilo Hertz).

Dans le mode de réalisation décrit et représenté ici, le signal de test x(n) est un signal de parole dépourvu de bruit. Le signal bruité xb(n) représente alors le signal vocal initial x(n) dégradé par un environnement bruité (bruit de fond ou bruit ambiant), et le signal y(n) représente le signal xb(n) après réduction de bruit.In the embodiment described and shown here, the test signal x (n) is a speech signal devoid of noise. The noisy signal xb (n) then represents the initial speech signal x (n) degraded by a noisy environment (background or ambient noise), and the signal y (n) represents the signal xb (n) after noise reduction.

Selon un exemple de mise en oeuvre de l'invention, le signal x(n) est généré dans une chambre anéchoïque. Cependant, le signal x(n) peut être aussi généré dans une pièce "calme" ayant un temps de réverbération "moyen", inférieur à 0,5 seconde.According to an exemplary implementation of the invention, the signal x (n) is generated in an anechoic chamber. However, the signal x (n) can also be generated in a "quiet" room having an "average" reverberation time of less than 0.5 seconds.

Le signal bruité xb(n) est obtenu en ajoutant une contribution prédéterminée de bruit au signal x(n). Le signal y(n) est obtenu soit en sortie d'un algorithme de réduction de bruit implanté sur un ordinateur personnel, soit à la sortie d'un équipement réseau réducteur de bruit et dans ce dernier cas, le signal y(n) est prélevé au niveau d'un codeur MIC (modulation par impulsion et codage).The noisy signal xb (n) is obtained by adding a predetermined contribution of noise to the signal x (n). The signal y (n) is obtained either at the output of a noise reduction algorithm implanted on a personal computer, or at the output of a noise reduction network equipment and in the latter case, the signal y (n) is taken at the level of a PCM encoder (pulse modulation and coding).

En référence à la figure 2 , le procédé de calcul de la note objective NOB_MOS de la gêne due au bruit dans le signal traité y(n) selon l'invention est représenté sous la forme d'un algorithme comportant des étapes a1 à a7.With reference to the figure 2 , the method of calculating the objective note NOB_MOS of the annoyance due to the noise in the processed signal y (n) according to the invention is represented in the form of an algorithm comprising steps a1 to a7.

Dans une première étape a1, les signaux x(n), xb(n) et y(n) sont respectivement découpés en fenêtres temporelles successives appelées trames. Chaque trame de signal, notée m, contient un nombre prédéterminé d'échantillons du signal, l'étape a1 consiste donc en un changement de cadence de chacun de ces signaux. Les signaux x(n), xb(n) et y(n) passés en cadence trames produisent respectivement les signaux x[m], xb[m], et y[m].In a first step a1, the signals x (n), xb (n) and y (n) are respectively divided into successive time windows called frames. Each signal frame, denoted m, contains a predetermined number of samples of the signal, step a1 therefore consists of a change in the rate of each of these signals. The signals x (n), xb (n) and y (n) in frame rate respectively produce the signals x [m], xb [m], and y [m].

Dans une seconde étape a2, une détection d'activité vocale (DAV) est effectuée sur le signal x[m] de manière à déterminer si chaque trame respective courante d'indice m des signaux xb[m] et y[m], est une trame contenant seulement du bruit, notée "m_bruit", ou une trame contenant de la parole, c'est-à-dire du signal utile, et notée "m_parole". Cette détermination se fait par comparaison des signaux xb[m] et y[m] avec le signal de test x[m] dénué de bruit. Chaque trame de silence de x[m] correspond en effet à une trame de bruit pour les signaux xb[m] et y[m], tandis que chaque trame de parole de x[m] correspond à une trame de parole pour les signaux xb[m] et y[m].In a second step a2, a speech activity detection (DAV) is performed on the signal x [m] so as to determine whether each respective current frame of index m of the signals xb [m] and y [m], is a frame containing only noise, denoted "m_noise", or a frame containing speech, that is to say the useful signal, and noted "m_parole". This determination is made by comparing the signals xb [m] and y [m] with the test signal x [m] devoid of noise. Each silence frame of x [m] corresponds in fact to a noise frame for the signals xb [m] and y [m], while each speech frame of x [m] corresponds to a speech frame for the signals xb [m] and y [m].

Comme représenté sur la figure 2 , en sortie de l'étape a2, trois types de trames sont sélectionnés à partir des signaux x[m], xb[m] et y[m] :

  • les trames de parole du signal bruité xb[m], notées xb[m_parole],
  • les trames de parole du signal traité y[m], notées y[m_parole],
  • les trames de bruit du signal traité y[m], notées y[m_bruit].
As shown on the figure 2 , At the output of step a2, three types of frames are selected from the signals x [m], xb [m] and y [m]:
  • speech frames of the noisy signal xb [m], denoted xb [m_parole],
  • the speech frames of the processed signal y [m], written y [m_parole],
  • the noise frames of the processed signal y [m], denoted y [m_noise].

Dans une troisième étape a3, des mesures de sonie sont effectuées sur au moins des ensembles de trames y[m_bruit], y[m_parole], xb[m_parole] issues de l'étape précédente a2, et au moins un ensemble de trames du signal y[m] en sortie de l'étape a1. Par exemple si on utilise 8 secondes de signal de test échantillonné à 8kHz, on pourra travailler sur 250 trames y[m] de 256 échantillons de signal y(n). De plus les coefficients de tonalité d'au moins un ensemble de trames y[m_bruit] sont mesurées.In a third step a3, loudness measurements are made on at least sets of y [m_noise], y [m_parole], xb [m_parole] frames from the previous step a2, and at least one set of frames of the signal y [m] at the output of step a1. For example, if 8 seconds of sampled test signal at 8 kHz is used, it will be possible to work on 250 fields y [m] of 256 samples of signal y (n). In addition, the tone coefficients of at least one set of y [m_noise] frames are measured.

Plus précisément, à cette étape, on calcule les densités de sonie moyennes S Xb (m_parole), S Y (m_parole), S Y (m), et S Y (m_bruit) de respectivement chacune des trames xb[m_parole], y[m_parole], y[m] et y[m_bruit] des ensembles de trames considérés. De même les coefficients de tonalité α Y (m_bruit) de chacune des trames y[m_bruit] de l'ensemble considéré de trames y[m_bruit] sont calculés.More specifically, at this stage, the mean loudness densities are calculated S Xb ( m_parole ), S Y ( m_parole ), S Y ( m ), and S Y ( m_noise ) of respectively each of the frames xb [m_parole], y [m_parole], y [m] and y [m_noise] of sets of frames considered. Similarly, the tone coefficients α Y ( m_noise ) of each of the y [m_noise] frames of the considered set of y [m_noise] frames are calculated.

Le calcul d'une densité de sonie moyenne S U (m) et d'un coefficient de tonalité α(m) d'une trame d'indice m quelconque d'un signal audio donné u, sera détaillé plus loin en liaison avec la figure 4 . Calculation of an average loudness S U (m) and a tone coefficient α (m) of a frame of any index m of a given audio signal u, will be detailed later in connection with the figure 4 .

Dans une quatrième étape a4, on calcule les moyennes respectives S Xb _ parole , S Y _ parole , S Y , et S Y _ bruit des densités de sonie moyenne S Xb (m _ parole), S Y (m _ parole), S Y (m), et S Y (m_bruit) précédemment calculées sur les ensembles respectifs considérés des trames xb[m_parole], y[m_parole], y[m] et y[m_bruit]. La moyenne α Y _ bruit des coefficients de tonalité α Y (m_bruit) précédemment calculés sur l'ensemble considéré de trames y[m_bruit] est également calculée.In a fourth step a4, the respective averages are calculated S Xb _ word , S Y _ word , S Y , and S Y _ noise of medium loudness densities S Xb ( m _ speech ), S Y ( m _ speech ), S Y ( m ), and S Y ( m _ noise ) previously calculated on the respective sets considered frames xb [m_parole], y [m_parole], y [m] and y [m_noise]. The average α Y _ noise α Y tone coefficients (m _ noise) previously calculated for all considered frames y [m_noise] is also calculated.

Dans une cinquième étape a5, on calcule cinq facteurs facteur(i), i étant un entier variant de un à cinq, caractéristiques de la gêne due au bruit dans le signal y(n), selon les formules suivantes:

  • facteur 1 = S Y _bruit S Y ,
    Figure imgb0010
  • facteur 2 = S Y _bruit S Y _parole ,
    Figure imgb0011
  • facteur(3)= Ecart_type ( S Xb (m_parole)- S Y (m_parole)), l'opérateur "Ecart_type (v(m))" désignant l'écart-type de la variable v sur l'ensemble des trames m,
  • facteur(4)= α Y _ bruit ,
  • facteur(5)= Ecart_type (α Y (m_bruit)).
In a fifth step a5, five factor factors (i) are calculated, i being an integer varying from one to five, characteristic of the annoyance due to the noise in the signal y (n), according to the following formulas:
  • postman 1 = S ~ Y _noise S ~ Y ,
    Figure imgb0010
  • postman 2 = S ~ Y _noise S ~ Y _parole ,
    Figure imgb0011
  • factor (3) = Standard deviation ( S Xb ( m_parole ) - S Y ( m_parole )), the operator "Ecart_type (v (m))" designating the standard deviation of the variable v on the set of frames m,
  • factor (4) = α Y _ noise ,
  • factor (5) = Standard deviation (α Y ( m_noise )).

Dans une sixième étape a6, le calcul d'une note objective intermédiaire NOB est obtenue par combinaison linéaire des cinq facteurs calculés à l'étape a5, suivant l'équation suivante: NOB = i = 1 5 ω i facteur i + ω 6 ,

Figure imgb0012

où les coefficients ω1 à ω6 sont des coefficients de pondération prédéfinis. Ces coefficients ont été déterminés de manière à obtenir une corrélation maximale entre les données subjectives issues d'une base de données de tests subjectifs, et les notes objectives NOB calculées par cette combinaison linéaire en utilisant les signaux de tests, bruités et traités x[m], xb[m] et y[m] utilisés lors de ces mêmes tests subjectifs. La base de données de tests subjectifs est par exemple une base de données de notes obtenues avec des groupes d'auditeurs conformément à la "Recommandation UIT-T P.835", dans laquelle ces notes sont appelées notes "background noise".In a sixth step a6, the calculation of an intermediate objective score NOB is obtained by linear combination of the five factors calculated in step a5, according to the following equation: NOB = Σ i = 1 5 ω i postman i + ω 6 ,
Figure imgb0012

where the coefficients ω 1 to ω 6 are predefined weighting coefficients. These coefficients were determined in order to obtain a maximum correlation between the subjective data from a subjective test database, and the objective scores NOB calculated by this linear combination using the test signals, noisy and processed x [m ], xb [m] and y [m] used in these same subjective tests. The subjective test database is for example a database of scores obtained with groups of listeners according to "ITU-T Recommendation P.835", in which these notes are called background noise notes.

Il est à noter que l'obtention des coefficients de pondération par l'utilisation d'une base de données de tests subjectifs n'est pas indispensable à chaque étape de calcul d'une note objective NOB. En effet, ces coefficients doivent être obtenus préalablement à la première utilisation du procédé, et peuvent être les mêmes pour toutes les utilisations du procédé. Ces coefficients sont néanmoins amenés à évoluer lorsque de nouvelles données subjectives viendront alimenter la base de données de tests subjectifs utilisée.It should be noted that the obtaining of the weighting coefficients by the use of a database of subjective tests is not essential for each step of calculating an objective score NOB. Indeed, these coefficients must be obtained prior to the first use of the process, and may be the same for all uses of the process. These coefficients are nevertheless likely to evolve when new subjective data come to feed the database of subjective tests used.

Enfin dans une dernière étape a7, une note objective NOB_MOS de la gêne due au bruit dans le signal traité y(n) sur l'échelle MOS est calculée en utilisant par exemple une fonction polynomiale d'ordre 3, suivant l'équation suivante: NOB_MOS = i = 1 4 λ i NOB i - 1 ,

Figure imgb0013

où les coefficients λ1 à λ4 sont déterminés de manière à ce que la note objective obtenue NOB_MOS caractérise la gêne due au bruit sur l'échelle MOS, c'est-à-dire sur une échelle de 1 à 5.Finally, in a last step a7, an objective note NOB_MOS of the annoyance due to the noise in the processed signal y (n) on the MOS scale is calculated using for example a polynomial function of order 3, according to the following equation: NOB_MOS = Σ i = 1 4 λ i NOB i - 1 ,
Figure imgb0013

where the coefficients λ 1 to λ 4 are determined so that the objective score obtained NOB_MOS characterizes the annoyance due to the noise on the MOS scale, that is to say on a scale of 1 to 5.

Selon un second mode de réalisation du procédé l'invention, la gêne due à la présence de bruit dans un signal audio bruité quelconque est évaluée de manière objective. On utilise le même environnement de test qu'à la figure 1 , mais en ôtant le module MRB de réduction de bruit. La source de signaux audio SSA délivre un signal audio de test x(n) ne contenant que du signal utile, auquel est ajouté un signal de bruit prédéfini généré par la source de bruit SB, pour obtenir en sortie de l'opérateur d'addition AD un signal bruité xb(n).According to a second embodiment of the method of the invention, the annoyance due to the presence of noise in any noisy audio signal is evaluated objectively. The same test environment is used as in the figure 1 , but by removing the MRB noise reduction module. The audio signal source SSA delivers a test audio signal x (n) containing only the wanted signal, to which is added a predefined noise signal generated by the noise source SB, to obtain at the output of the addition operator AD a noisy signal xb (n).

Le signal de test x(n) et le signal bruité xb(n) sont alors directement envoyés à l'entrée de l'équipement de test EQT mettant en oeuvre un procédé d'évaluation objective de la gêne due au bruit dans le signal bruité xb(n) selon l'invention. Comme dans le premier mode de réalisation, les signaux x(n) et xb(n) sont supposés échantillonnés à la fréquence d'échantillonnage 8 kHz.The test signal x (n) and the noisy signal xb (n) are then directly sent to the input of the test equipment EQT implementing a method of objective evaluation of the annoyance due to the noise in the noisy signal. xb (n) according to the invention. As in the first embodiment, the signals x (n) and xb (n) are assumed to be sampled at the 8 kHz sampling rate.

L'équipement de test EQT délivre en sortie un résultat d'évaluation RES, qui est une note objective NOB_MOS de la gêne due à la présence de bruit dans le signal bruité xb(n).The test equipment EQT outputs an evaluation result RES, which is an objective note NOB_MOS of the annoyance due to the presence of noise in the noisy signal xb (n).

En référence à la figure 3 , le procédé de calcul de la note objective NOB_MOS de la gêne due au bruit dans le signal bruité xb(n) selon l'invention est représenté sous la forme d'un algorithme comportant des étapes b1 à b7. Ces étapes sont similaires aux étapes a1 à a7 précédemment décrites dans le premier mode de réalisation, et seront donc un peu moins détaillées. Il est en effet à noter que si l'on applique les étapes de calcul a3 à a7 avec le signal y(n) égal au signal xb(n) dans le cas du premier mode de réalisation, on aboutit au deuxième mode de réalisation.With reference to the figure 3 , the method for calculating the objective note NOB_MOS of the annoyance due to the noise in the noisy signal xb (n) according to the invention is represented in the form of an algorithm comprising steps b1 to b7. These steps are similar to steps a1 to a7 previously described in the first embodiment, and will therefore be a little less detailed. It should be noted that if we apply the calculation steps a3 to a7 with the signal y (n) equal to the signal xb (n) in the case of the first embodiment, we reach the second embodiment.

Dans une première étape b1, les signaux x(n) et xb(n) sont découpés en trames x[m] et xb[m] d'indice temporel m.In a first step b1, the signals x (n) and xb (n) are split into frames x [m] and xb [m] of time index m.

Dans une seconde étape b2, une détection d'activité vocale est effectuée sur le signal x[m] de manière à déterminer si chaque trame courante d'indice m du signal bruité xb[m] est une trame contenant seulement du bruit, notée "m_bruit", ou une trame contenant aussi de la parole, notée "m_parole". Deux types de trames sont donc sélectionnés à partir des signaux x[m] et xb[m] en sortie de l'étape b2:

  • les trames de parole du signal bruité xb[m], notées xb[m_parole],
  • et les trames de bruit du signal bruité xb[m], notées xb[m_bruit].
In a second step b2, a voice activity detection is performed on the signal x [m] so as to determine whether each current frame of index m of the noisy signal xb [m] is a frame containing only noise, denoted " m_noise ", or a frame also containing speech, denoted" m_parole ". Two types of frames are thus selected from the signals x [m] and xb [m] at the output of step b2:
  • speech frames of the noisy signal xb [m], denoted xb [m_parole],
  • and the noise frames of the noisy signal xb [m], denoted xb [m_noise].

Dans une troisième étape b3, des mesures de sonie sont effectuées sur au moins des ensembles de trames xb[m_bruit] et xb[m_parole] issues de l'étape précédente b2, et au moins un ensemble de trames du signal xb[m] en sortie de l'étape b1. De plus les coefficients de tonalité d'au moins un ensemble de trames xb[m_bruit] sont mesurées.In a third step b3, loudness measurements are made on at least sets of frames xb [m_noise] and xb [m_parole] from the previous step b2, and at least one set of frames of the signal xb [m] in exit from step b1. In addition, the tone coefficients of at least one set of frames xb [m_noise] are measured.

Plus précisément, à cette étape, on calcule les densités de sonie moyennes S Xb (m). S Xb (m_parole) et S Xb (m_bruit) de respectivement chacune des trames xb[m], xb[m_parole] et xb[m_bruit] des ensembles de trames considérés. De même les coefficients de tonalité α Xb (m_bruit) de chacune des trames xb[m_bruit] de l'ensemble considéré de trames xb[m_bruit] sont calculés.More specifically, at this stage, the mean loudness densities are calculated S Xb ( m ). S Xb ( m _ word ) and S Xb ( m_noise ) respectively of the frames xb [m], xb [m_parole] and xb [m_noise] of the sets of frames considered. Similarly, the tone coefficients α Xb ( m_noise ) of each frames xb [m_noise] of the considered set of frames xb [m_noise] are calculated.

Dans une quatrième étape b4, on calcule les moyennes respectives S Xb , S Xb _ parole, et S Xb _ bruit des densités de sonie moyenne S Xb (m), S Xb (m_parole) et S Xb (m_bruit) précédemment calculées sur les ensembles respectifs considérés des trames xb[m], xb[m_parole] et xb[m_bruit]. La moyenne α Xb _ bruit des coefficients de tonalité α Xb (m_bruit) précédemment calculés sur l'ensemble considéré de trames xb[m_bruit] est également calculée.In a fourth step b4, the respective averages are calculated S Xb , S Xb _ word, and S Xb _ noise of medium loudness densities S Xb ( m ), S Xb ( m_parole ) and S Xb ( m_noise ) previously calculated on the respective sets considered frames xb [m], xb [m_parole] and xb [m_noise]. The mean α Xb _ noise of the tone coefficients α Xb ( m_noise ) previously calculated on the considered set of frames xb [m_noise] is also calculated.

Dans une cinquième étape b5, on calcule quatre facteurs facteur(i), i étant un entier variant de un à quatre, caractéristiques de la gêne due au bruit dans le signal bruité xb(n), selon les formules suivantes:

  • facteur 1 = S Xb _bruit S Xb ,
    Figure imgb0014
  • facteur 2 = S Xb _bruit S Xb _parole ,
    Figure imgb0015
  • facteur(3)= αXb_bruit,
  • facteur(3)= α Xb _ bruit ,
  • facteur(4)= Ecart_type(α Xb (m_bruit)), l'opérateur "Ecart_type (v(m))" désignant l'écart-type de la variable v sur l'ensemble des trames m.
In a fifth step b5, four factor factors (i) are calculated, i being an integer varying from one to four, characteristic of the annoyance due to the noise in the noisy signal xb (n), according to the following formulas:
  • postman 1 = S ~ Xb _noise S ~ Xb ,
    Figure imgb0014
  • postman 2 = S ~ Xb _noise S ~ Xb _parole ,
    Figure imgb0015
  • factor (3) = α Xb_noise ,
  • factor (3) = α Xb _ noise ,
  • factor (4) = Standard deviation (α Xb ( m_noise )), the operator "Ecart_type (v (m))" designating the standard deviation of the variable v over the set of frames m.

Dans une sixième étape b6, le calcul d'une note objective intermédiaire NOB est obtenue par combinaison linéaire des quatre facteurs calculés à l'étape b5, suivant l'équation suivante: NOB = i = 1 4 ω i facteur i + ω 5 ,

Figure imgb0016

où les coefficients ω1 à ω5 sont des coefficients de pondération prédéfinis. Ces coefficients ont été déterminés de manière à obtenir une corrélation maximale entre les données subjectives issues d'une base de données de tests subjectifs, et les notes objectives NOB calculées par cette combinaison linéaire en utilisant les signaux de tests et les signaux bruités x[m] et xb[m] utilisés lors de ces mêmes tests subjectifs. Tout comme pour l'étape a6, l'obtention des coefficients de pondération par l'utilisation d'une base de données de tests subjectifs n'est pas indispensable à chaque étape de calcul d'une note objective NOB.In a sixth step b6, the calculation of an intermediate objective score NOB is obtained by linear combination of the four factors calculated in step b5, according to the following equation: NOB = Σ i = 1 4 ω i postman i + ω 5 ,
Figure imgb0016

where the coefficients ω 1 to ω 5 are predefined weighting coefficients. These coefficients were determined in order to obtain a maximum correlation between the subjective data from a subjective test database, and the objective scores NOB calculated by this linear combination using the test signals and the noisy signals x [m ] and xb [m] used in these same subjective tests. Just as for step a6, obtaining the weighting coefficients by the use of a database of subjective tests is not indispensable at each step of calculating an objective score NOB.

Enfin dans une dernière étape b7, une note objective NOB_MOS de la gêne due au bruit dans le signal bruité xb(n) sur l'échelle MOS est calculée en utilisant par exemple une fonction polynomiale d'ordre 3, suivant l'équation suivante: NOB_MOS = i = 1 4 λ i NOB i - 1 ,

Figure imgb0017

où les coefficients λ1 à λ4 sont déterminés de manière à ce que la note objective obtenue NOB_MOS caractérise la gêne due au bruit sur l'échelle MOS, c'est-à-dire sur une échelle de 1 à 5.Finally, in a last step b7, an objective note NOB_MOS of the annoyance due to the noise in the noisy signal xb (n) on the MOS scale is calculated using for example a polynomial function of order 3, according to the following equation: NOB_MOS = Σ i = 1 4 λ i NOB i - 1 ,
Figure imgb0017

where the coefficients λ 1 to λ 4 are determined so that the objective score obtained NOB_MOS characterizes the annoyance due to the noise on the MOS scale, that is to say on a scale of 1 to 5.

Le calcul de densité de sonie moyenne et du coefficient de tonalité d'une trame d'un signal audio, utilisé dans les étapes a3 et b3, est maintenant décrit en relation avec la figure 4 , selon un mode de réalisation préféré de l'invention.The calculation of mean loudness density and tone coefficient of a frame of an audio signal, used in steps a3 and b3, is now described in relation to the figure 4 According to a preferred embodiment of the invention.

Le calcul selon l'invention de la densité de sonie moyenne S U(m) d'une trame d'indice m quelconque d'un signal audio donné u[m], comprend les étapes c1 à c7 représentées à la figure 4 et exposées ci-après. Le calcul selon l'invention du coefficient de tonalité α(m) d'une trame d'indice m quelconque d'un signal audio donné u[m], comprend les étapes c1, c2, c3 et c8 représentées à la figure 4 et exposées ci-après.The calculation according to the invention of the average loudness density S U (m) of a frame of any index m of a given audio signal u [m], comprises the steps c1 to c7 represented in FIG. figure 4 and explained below. The calculation according to the invention of the tone coefficient α (m) of a frame of any index m of a given audio signal u [m], comprises the steps c1, c2, c3 and c8 represented in FIG. figure 4 and explained below.

Dans ce qui suit, on considère une trame d'indice m quelconque d'un signal u[m], sachant que tout ou partie des trames du signal considéré subissent le même traitement. Le signal u[m] représente n'importe lequel des signaux x[m], xb[m], ou y[m] définis plus haut.In what follows, we consider a frame of any index m of a signal u [m], knowing that all or part of the frames of the signal considered undergo the same treatment. The signal u [m] represents any of the signals x [m], xb [m], or y [m] defined above.

A la première étape c1, on applique à la trame d'indice m du signal u[m] un fenêtrage, par exemple un fenêtrage de type Hanning, Hamming ou équivalent. On obtient alors une trame fenêtrée u_w[m].In the first step c1, a windowing is applied to the frame of index m of the signal u [m], for example a windowing of Hanning, Hamming or equivalent type. We then obtain a windowed frame u_w [m].

A l'étape suivante c2, on applique à la trame fenêtrée u_w[m], une transformée de Fourier rapide (FFT) et on obtient en conséquence une trame correspondante U(m,f) dans le domaine fréquentiel.In the next step c2, a fast Fourier transform (FFT) is applied to the windowed frame u_w [m] and a corresponding frame U (m, f) in the frequency domain is accordingly obtained.

A l'étape suivante c3, on calcule la densité spectrale de puissance γU(m, f) de la trame U(m,f). Un tel calcul est connu de l'homme du métier et ne sera pas, par conséquent, détaillé ici.In the next step c3, the power spectral density γ U (m, f) of the frame U (m, f) is calculated. Such a calculation is known to those skilled in the art and will not, therefore, be detailed here.

A l'issue de l'étape c3, pour le signal y[m_bruit] de l'étape a3 ou le signal xb[m_bruit] de l'étape b3, on passe par exemple à l'étape c8 pour le calcul du coefficient de tonalité, puis à l'étape c4 pour le calcul de la densité de sonie moyenne S U(m), puisque pour ces deux signaux les deux calculs sont nécessaires. Pour les autres signaux des étapes a3 et b3 on passe à l'étape c4 pour le calcul de la densité de sonie moyenne S U(m). Il est à noter que le calcul du coefficient de tonalité est indépendant du calcul de la densité de sonie moyenne S U(m), les deux calculs peuvent donc s'effectuer en parallèle ou l'un après l'autre.At the end of step c3, for the signal y [m_noise] of step a3 or the signal xb [m_noise] of step b3, for example, step c8 is used to calculate the coefficient of tone, then at step c4 for calculating the average loudness of loudness S U (m), since for these two signals the two calculations are necessary. For the other signals of steps a3 and b3, we go to step c4 for the calculation of the average loudness density S U (m). It should be noted that the calculation of the tone coefficient is independent of the calculation of the mean loudness density S U (m), the two calculations can be carried out in parallel or one after the other.

A l'étape c4, on applique à la densité spectrale de puissance γU(m, f) obtenue à l'étape précédente, une conversion de l'axe des fréquences à l'échelle des Barks, et on obtient en conséquence une densité spectrale de puissance, BU(m, b), sur l'échelle des Barks, appelée aussi spectre de Bark. Pour une fréquence d'échantillonnage de 8kHz, 18 bandes critiques doivent être considérées. Ce type de conversion est connu de l'homme du métier, le principe de cette conversion Hertz/Bark consiste à additionner toutes les contributions fréquentielles présentes dans la bande critique considérée de l'échelle des Barks.In step c4, a frequency conversion of the frequency axis at the Barks scale is applied to the power spectral density γ U (m, f) obtained in the previous step, and a density is consequently obtained. spectral power, B U (m, b), on the Barks scale, also called Bark spectrum. For a sampling frequency of 8kHz, 18 critical bands must be considered. This type of conversion is known to those skilled in the art, the principle of this Hertz / Bark conversion is to add all the frequency contributions present in the critical band considered Barks scale.

Ensuite, à l'étape c5, on applique à la densité spectrale de puissance sur l'échelle des Barks, BU(m, b), une convolution avec la fonction d'étalement couramment utilisée en psychoacoustique, et on obtient en conséquence une densité spectrale étalée sur l'échelle des Barks, notée EU(m, b). Cette fonction d'étalement a été formulée mathématiquement et une expression possible est: 10 log 10 E b = 15.81 + 7.5 * b + 0.474 - 17.5 * 1 + b + 0.474 2 ,

Figure imgb0018

où E(b) est la fonction d'étalement appliquée à la bande critique b considérée dans l'échelle des Barks et * symbolise l'opérateur de multiplication dans l'espace des nombres réels. Cette étape permet de prendre en compte l'interaction des bandes critiques adjacentes.Then, in step c5, the power spectral density on the Barks scale, B U (m, b), is subjected to a convolution with the spreading function commonly used in psychoacoustics, and a result is consequently obtained. spectral density spread over the Barks scale, denoted E U (m, b). This spreading function has been mathematically formulated and one possible expression is: 10 log 10 E b = 15.81 + 7.5 * b + 0474 - 17.5 * 1 + b + 0474 2 ,
Figure imgb0018

where E (b) is the spread function applied to the critical band b considered in the Barks scale and * symbolizes the multiplication operator in the real number space. This step makes it possible to take into account the interaction of the adjacent critical bands.

A l'étape suivante c6, on convertit la densité spectrale étalée EU(m, b) obtenue précédemment en densités de sonie exprimées en sones. Pour cela, on opère une calibration de la densité spectrale étalée sur l'échelle des Barks, EU(m, b), par les facteurs respectifs d'échelonnement en puissance et d'échelonnement en sonie couramment utilisés en psychoacoustique. Le document "Recommandation UIT-T P.862", sections 10.2.1.3 et 10.2.1.4, donne un exemple d'une telle calibration par les facteurs précités. On convertit ensuite sur l'échelle des phones la grandeur obtenue. La conversion sur l'échelle des phones est effectuée en s'appuyant sur les courbes d'isosonie (courbes de Fletcher) conformément à la norme NF ISO 226 "Lignes isosoniques normales". On effectue alors une conversion sur l'échelle des sones de la grandeur précédemment convertie en phones. La conversion en sones est effectuée conformément à la loi de Zwicker selon laquelle : N sone = 2 N phone - 40 10

Figure imgb0019
In the next step c6, the spread spectrum density E U (m, b) obtained previously is converted into loudness densities expressed in sones. For this, a calibration of the spectral density spread on the Barks scale, E U (m, b), is performed by the respective power scaling and loudness scaling factors commonly used in psychoacoustics. The document "ITU-T Recommendation P.862", sections 10.2.1.3 and 10.2.1.4, gives an example of such a calibration by the aforementioned factors. The size obtained is then converted on the scale of the phones. The conversion on the scale of the phones is carried out based on the isosonic curves (Fletcher curves) in accordance with the standard NF ISO 226 "Normal isosonic lines". We then perform a conversion on the scale of sones of the size previously converted to phones. The conversion to sones is made in accordance with Zwicker's law that: NOT sone = 2 NOT phone - 40 10
Figure imgb0019

Pour obtenir plus d'information sur la conversion phone/sone, on pourra se reporter au document "PSYCHOACOUSTIQUE, L'oreille récepteur d'information", de E. Zwicker et R. Feldtkeller, édition Masson, 1981.For more information on the conversion of phone / sone, see the document "PSYCHOACOUSTIQUE, The ear receiver of information", by E. Zwicker and R. Feldtkeller, Masson edition, 1981.

A l'issue de l'étape c6, on dispose d'un nombre B de valeurs de densité de sonie, SU(m, b), de la trame d'indice m pour la bande critique b, B étant le nombre de bandes critiques considérées dans l'échelle des Barks et l'indice b variant de 1 à B.At the end of step c6, there is a number B of loudness density values, S U (m, b), of the frame of index m for the critical band b, where B is the number of critical bands considered in the Barks scale and the index b varying from 1 to B.

Enfin, à l'étape c7, on calcule la densité de sonie moyenne S U(m) de la trame d'indice m à partir desdites B valeurs de densité de sonie, selon l'équation suivante : S U m = 1 B b = 1 B S U m b

Figure imgb0020
Finally, in step c7, the mean loudness density is calculated S U (m) of the frame of index m from said B loudness density values, according to the following equation: S ~ U m = 1 B Σ b = 1 B S U m b
Figure imgb0020

Autrement dit, la densité de sonie moyenne S U(m) selon l'invention d'une trame d'indice m, est donc la moyenne des B valeurs de densité de sonie SU(m, b), de la trame d'indice m pour une bande critique b considérée.In other words, the average loudness of loudness S U (m) according to the invention of a frame of index m, is therefore the average of the B loudness density values S U (m, b), of the frame of index m for a critical band b considered.

Ces deux dernières étapes c6 et c7 correspondent à une conversion du domaine des Barks vers le domaine des Sones, permettant de calculer une intensité subjective moyenne, c'est-à-dire telle que perçue par l'oreille humaine.These last two steps c6 and c7 correspond to a conversion of the Barks domain to the Sones domain, making it possible to calculate a mean subjective intensity, that is to say as perceived by the human ear.

En outre à l'étape c8, le coefficient de tonalité α(m) de la trame d'indice m est calculé selon l'équation suivante: α m = 10 * log 10 f = 0 N - 1 γ U m f 1 / N 1 N f = 0 N - 1 γ U m f - 60 ,

Figure imgb0021

où * symbolise l'opérateur de multiplication dans l'espace des nombres réels, f représente l'indice fréquentiel de la densité spectrale de puissance, et N désigne la taille de la transformée de Fourier rapide. Ce calcul est effectué selon le principe défini par J.D. Johnston dans son article " Transform coding of audio signais using perceptual noise criteria" du journal "IEEE Journal on selected areas in communications, vol.6, n°2, February 1988 ".In addition, in step c8, the tone coefficient α (m) of the frame of index m is calculated according to the following equation: α m = 10 * log 10 Π f = 0 NOT - 1 γ U m f 1 / NOT 1 NOT Σ f = 0 NOT - 1 γ U m f - 60 ,
Figure imgb0021

where * symbolizes the multiplication operator in the real number space, f represents the frequency index of the power spectral density, and N denotes the size of the fast Fourier transform. This calculation is done according to the principle defined by JD Johnston in his article " Transform coding of audio using the perceptual noise criteria of the journal IEEE Journal on selected areas in communications, vol. 6, no. 2, February 1988 ".

Le coefficient de tonalité α d'un signal de base est une mesure permettant de montrer si certaines fréquences pures ressortent de ce signal. Il est équivalent à une densité tonale. En effet, plus le coefficient de tonalité α est proche de 0, plus le signal est assimilé à du bruit. A l'inverse, plus le coefficient de tonalité α est proche de 1, plus le signal est à composante tonale majoritaire. Un coefficient de tonalité α proche de 1 atteste donc de la présence de signal utile, ou signal de parole.The tone coefficient α of a basic signal is a measure to show whether certain pure frequencies emerge from this signal. It is equivalent to a tonal density. Indeed, the more the tone coefficient α is close to 0, the more the signal is likened to noise. Conversely, the more the tone coefficient α is close to 1, the more the signal is component tonal majority. A tone coefficient α close to 1 attests to the presence of useful signal, or speech signal.

Claims (13)

  1. Method of calculating an objective score (NOB) of the nuisance caused by noise in an audio signal processed by a noise-reducing function, said method comprising a preliminary step of obtaining a predefined test audio signal (x[m]) containing a wanted signal without any noise, a noise-affected signal (xb[m]), obtained by adding a predefined noise signal to said test signal (x[m]), and a processed signal (y[m]), obtained by application of the noise-reducing function to said noise-affected signal (xb[m]), said method being characterized in that it includes a step (a3, a4):
    - of calculating loudness densities of frames of said noise-affected signal (xb[m]) and of said processed signal (y[m]), said loudness densities for any frame m of a given signal u[m] being obtained from the spread spectral density on the Barks scale, EU(m, b), of the signal u[m], by an operation of calibrating the spread spectral density by respective power and loudness grading factors, followed by a conversion operation on the phons scale and on the sones scale; and
    - of calculating tone coefficients of frames of said processed signal (y[m]), the tone coefficient, α(m), of any frame of index m of a given signal u[m] being calculated according to the following equation: α m = 10 * log 10 f = 0 N - 1 γ U m f 1 / N 1 N f = 0 N - 1 γ U m f - 60
    Figure imgb0033
    in which γU(m, f) designates the power spectral density obtained for any frame m of the signal u[m].
  2. Method according to Claim 1, characterized in that it comprises the steps of:
    - calculation (a3) of average loudness densities S Y(m) of frames of the processed signal (y[m]), of respective average loudness densities S Xb(m_speech) and S Y(m_speech) of frames of wanted signal "m_speech" respectively of the noise-affected signal (xb[m]) and of the processed signal (y[m]), of average loudness densities S Y(m_noise) of noise frames "m_noise" of the processed signal (y[m]), and of tone coefficients αy(m_noise) of noise frames "m_noise" of the processed signal (y[m]),
    - calculation (a5, a6) of an objective score (NOB) of the nuisance due to the noise in the processed signal (y[m]), from said average loudness densities and said calculated tone coefficients, and of predefined weighting coefficients.
  3. Method according to Claim 2, characterized in that the step (a3) of calculating average loudness densities and tone coefficients is followed by a step (a4) of calculating the averages S Y, S Xb_speech, S Y_speech, S Y_noise and αY_noise of said average loudness densities and of said tone coefficients over all the relevant frames of the corresponding signals, and in that the objective score (NOB) of the nuisance due to the noise is calculated according to the following equation: NOB = i = 1 5 ω i factor i + ω 6 ,
    Figure imgb0034

    in which
    factor 1 = S Y _noise S Y ,
    Figure imgb0035

    factor 2 = S Y _noise S Y _speech ,
    Figure imgb0036

    factor (3) = standard_deviation (S Xb (m_speech)-S Y(m_speech)), the operator "standard_deviation (v(m))" designating the standard deviation of the variable v over all the frames of index m,
    factor (4) = αY_noise,
    factor (5) = standard_deviation(αY(m_noise)),
    and the coefficients ω1 to ω6 are determined in such a way as to obtain a maximum correlation between the subjective data obtained from a subjective test database and the objective scores (NOB) calculated by said method for the corresponding test, noise-affected and processed signals (x[m], xb[m], y[m]) used on said subjective tests.
  4. Method of calculating an objective score (NOB) of the nuisance due to the noise in an audio signal, said method comprising a preliminary step of obtaining a predefined test audio signal (x[m]) containing a wanted signal without noise, and a noise-affected signal (xb[m]), obtained by adding a predefined noise signal to said test signal (x[m]), said method being characterized in that it includes a step (b3, b4):
    - of calculating loudness densities of frames of said noise-affected signal (xb[m]), said loudness densities for any frame m of a given signal u[m] being obtained from the spread spectral density on the Barks scale, EU(m, b), of the signal u[m], by an operation for calibrating the spread spectral density by respective power and loudness grading factors, followed by a conversion operation on the phons scale and on the sones scale; and
    - of calculating tone coefficients of frames of said noise-affected signal (xb[m]), the tone coefficient, α(m), of any frame of index m of a given signal u[m] being calculated according to the following equation: α m = 10 * log 10 f = 0 N - 1 γ U m f 1 / N 1 N f = 0 N - 1 γ U m f - 60
    Figure imgb0037
    in which γU(m,f) designates the power spectral density obtained for any frame m of the signal u[m].
  5. Method according to Claim 4, characterized in that it comprises the steps of:
    - calculating (b3) average loudness densities S Xb(m) of frames of the noise-affected signal (xb[m]), average loudness densities S Xb(m_speech) of frames of wanted signal "m_speech" of the noise-affected signal (xb[m]), average loudness densities S Xb(m_noise) of noise frames "m_noise" of the noise-affected signal (xb[m]) and tone coefficients αXb(m_noise) of noise frames "m_noise" of the noise-affected signal (xb[m]),
    - calculating (b5, b6) an objective score (NOB) of the nuisance due to the noise in the noise-affected signal (xb[m]), from said average loudness densities and from said calculated tone coefficients, and predefined weighting coefficients.
  6. Method according to Claim 5, characterized in that the step (b3) of calculating average loudness densities and tone coefficients is followed by a step (b4) of calculating the averages S Xb, S Xb_speech, S Xb_noise and αXb_noise of said average loudness densities and of said tone coefficients over all the relevant frames of the corresponding signals, and in that said objective score (NOB) of the nuisance due to the noise is calculated according to the following equation: NOB = i = 1 4 ω i factor i + ω 5 ,
    Figure imgb0038

    in which
    factor 1 = S Xb _noise S Xb ,
    Figure imgb0039

    factor 2 = S Xb _noise S Xb _speech ,
    Figure imgb0040

    factor (3) = αXb_noise,
    factor (4) = standard_deviation (αXb(m_noise)), the operator "standard_deviation (v(m))" designating the standard deviation of the variable v over all the frames of index m,
    and the coefficients ω1 to ω5 are determined in such a way as to obtain a maximum correlation between the subjective data obtained from a subjective test database and the objective scores (NOB) calculated by said method for the test signals and the corresponding noise-affected signals (x[m], xb[m]) used on said subjective tests.
  7. Method according to any one of Claims 1 to 6, characterized in that said step (a3, b3, a4, b4) of calculating loudness densities and tone coefficients is preceded by a step (a2, b2) of detecting voice activity on the test signal, so as to determine whether a current frame of index m of the noise-affected signal (xb[m]), and of the processed signal (y[m]) in the case of Claims 1 to 3, is a frame "m_noise" containing only noise, or a frame "m_speech" containing speech, called wanted signal frame.
  8. Method according to any one of Claims 1 to 7, characterized in that the step (a6, b6) of calculating the objective score (NOB) is followed by a step (a7, b7) of calculating an objective score on the MOS scale (NOB_MOS) of the nuisance due to the noise, calculated according to the following equation: NOB_MOS = i = 1 4 λ i NOB i - 1 ,
    Figure imgb0041

    in which the coefficients λ1 to λ4 are determined in such a way that said new objective score obtained (NOB_MOS) characterizes the nuisance due to the noise on the MOS scale.
  9. Method according to any one of Claims 1 to 8, characterized in that, in the step (a3, b3, a4, b4) of calculating loudness densities and tone coefficients, the calculation of the average loudness density S U(m) of any frame of index m of a given audio signal u, comprises the following steps:
    - windowing (c1), for example by the Hanning method, the frame of index m and obtaining a windowed frame u_w[m],
    - applying (c2) a fast Fourier transform to the windowed frame u_w[m] and obtaining a corresponding frame U(m, f) in the frequency domain,
    - calculating (c3) the power spectral density (γU(m,f) of the frame U(m,f),
    - applying (c4) to the power spectral density γU(m,f) a conversion of the axis of the frequencies to the Barks scale and obtaining a power spectral density BU(m, b) on the Barks scale,
    - convoluting (c5) the power spectral density on the Barks scale, BU(m, b), with the spreading function commonly used in psycho-acoustics and obtaining a spread spectral density on the Barks scale, EU(m,b),
    - calibrating (c6) the spread spectral density on the Barks scale, EU(m, b), by the respective power grading and loudness grading factors commonly used in psycho-acoustics, converting the duly obtained quantity to the phons scale then converting the quantity previously converted into phons to the sones scale, and consequently obtaining a number B of loudness density values, SU(m, b), of the frame of index m for the critical band b, B being the number of critical bands concerned in the Barks scale and the index b varying from 1 to B,
    - calculating (c7) the average loudness density S U(m) of the frame of index m from said B loudness density values SU(m, b), according to the following equation: S U m = 1 B b = 1 B S U m b
    Figure imgb0042
  10. Method according to any one of Claims 1 to 9, characterized in that, in the step (a3, b3, a4, b4) of calculating loudness densities and tone coefficients, the calculation of the tone coefficient α(m) of any frame of index m of a given audio signal u comprises the following steps:
    - windowing (c1), for example by the Hanning method, the frame of index m and obtaining a windowed frame u_w[m],
    - applying (c2) a fast Fourier transform to the windowed frame u_w[m] and obtaining a corresponding frame U(m,f) in the frequency domain,
    - calculating (c3) the power spectral density γU(m,f) of the frame U(m,f),
    - calculating (c8) the tone coefficient α(m) according to the following equation: α m = 10 * log 10 f = 0 N - 1 γ U m f 1 / N 1 N f = 0 N - 1 γ U m f - 60
    Figure imgb0043
    in which * symbolizes the multiplication operator in the space of the real numbers, f represents the frequency index of the power spectral density and N designates the size of the fast Fourier transform.
  11. Test equipment intended to assess an objective score of the nuisance due to the noise in an audio signal, characterized in that it comprises means suitable for implementing a method according to any one of Claims 1 to 10.
  12. Test equipment according to Claim 11,
    characterized in that it includes computer means and a computer program, said program comprising instructions suitable for implementing said method, when it is run by said computer means.
  13. Computer program on a computer medium, characterized in that it comprises instructions suitable for implementing a method according to any one of Claims 1 to 10, when the program is loaded and run in a computer system.
EP06709505A 2005-02-18 2006-02-13 Method of measuring annoyance caused by noise in an audio signal Not-in-force EP1849157B1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
FR0501747A FR2882458A1 (en) 2005-02-18 2005-02-18 METHOD FOR MEASURING THE GENE DUE TO NOISE IN AN AUDIO SIGNAL
PCT/FR2006/050126 WO2006087490A1 (en) 2005-02-18 2006-02-13 Method of measuring annoyance caused by noise in an audio signal

Publications (2)

Publication Number Publication Date
EP1849157A1 EP1849157A1 (en) 2007-10-31
EP1849157B1 true EP1849157B1 (en) 2009-07-29

Family

ID=34981381

Family Applications (1)

Application Number Title Priority Date Filing Date
EP06709505A Not-in-force EP1849157B1 (en) 2005-02-18 2006-02-13 Method of measuring annoyance caused by noise in an audio signal

Country Status (7)

Country Link
US (1) US20080267425A1 (en)
EP (1) EP1849157B1 (en)
AT (1) ATE438173T1 (en)
DE (1) DE602006008111D1 (en)
ES (1) ES2329932T3 (en)
FR (1) FR2882458A1 (en)
WO (1) WO2006087490A1 (en)

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102005052630A1 (en) * 2005-08-25 2007-03-01 Robert Bosch Gmbh Method and device for assessing the annoyance of squeaking noises
KR100810077B1 (en) * 2006-05-26 2008-03-05 권대훈 Equaliztion Method with Equal Loudness Curve
EP2374211B1 (en) 2008-12-24 2012-04-04 Dolby Laboratories Licensing Corporation Audio signal loudness determination and modification in the frequency domain
EP2685448B1 (en) * 2012-07-12 2018-09-05 Harman Becker Automotive Systems GmbH Engine sound synthesis
ES2613651T3 (en) 2013-01-29 2017-05-25 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Quantification of adaptive audio signals by low complexity tone
WO2017218989A1 (en) 2016-06-17 2017-12-21 Predictive Safety Srp, Inc. Impairment detection system and method
CN110688712A (en) * 2019-10-11 2020-01-14 湖南文理学院 Evaluation index for objective annoyance degree of automobile wind vibration noise sound quality and calculation method thereof
CN113473314A (en) * 2020-03-31 2021-10-01 华为技术有限公司 Audio signal processing method and related device
CN116429245B (en) * 2023-06-13 2023-09-01 江铃汽车股份有限公司 Method and system for testing noise of wiper motor

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5574824A (en) * 1994-04-11 1996-11-12 The United States Of America As Represented By The Secretary Of The Air Force Analysis/synthesis-based microphone array speech enhancer with variable signal distortion
FI100840B (en) * 1995-12-12 1998-02-27 Nokia Mobile Phones Ltd Noise attenuator and method for attenuating background noise from noisy speech and a mobile station
US6446038B1 (en) * 1996-04-01 2002-09-03 Qwest Communications International, Inc. Method and system for objectively evaluating speech
EP0980064A1 (en) * 1998-06-26 2000-02-16 Ascom AG Method for carrying an automatic judgement of the transmission quality of audio signals
FI114833B (en) * 1999-01-08 2004-12-31 Nokia Corp A method, a speech encoder and a mobile station for generating speech coding frames
US6490552B1 (en) * 1999-10-06 2002-12-03 National Semiconductor Corporation Methods and apparatus for silence quality measurement
FI116643B (en) * 1999-11-15 2006-01-13 Nokia Corp Noise reduction
DE60104091T2 (en) * 2001-04-27 2005-08-25 CSEM Centre Suisse d`Electronique et de Microtechnique S.A. - Recherche et Développement Method and device for improving speech in a noisy environment
US7590530B2 (en) * 2005-09-03 2009-09-15 Gn Resound A/S Method and apparatus for improved estimation of non-stationary noise for speech enhancement

Also Published As

Publication number Publication date
ATE438173T1 (en) 2009-08-15
US20080267425A1 (en) 2008-10-30
EP1849157A1 (en) 2007-10-31
ES2329932T3 (en) 2009-12-02
WO2006087490A1 (en) 2006-08-24
FR2882458A1 (en) 2006-08-25
DE602006008111D1 (en) 2009-09-10

Similar Documents

Publication Publication Date Title
EP1849157B1 (en) Method of measuring annoyance caused by noise in an audio signal
EP2419900B1 (en) Method and device for the objective evaluation of the voice quality of a speech signal taking into account the classification of the background noise contained in the signal
EP2415047B1 (en) Classifying background noise contained in an audio signal
EP1356461B1 (en) Noise reduction method and device
EP1593116B1 (en) Method for differentiated digital voice and music processing, noise filtering, creation of special effects and device for carrying out said method
US20150154978A1 (en) \system and method for noise reduction in processing speech signals by targeting speech and disregarding noise
CN104919525B (en) For the method and apparatus for the intelligibility for assessing degeneration voice signal
EP2772916B1 (en) Method for suppressing noise in an audio signal by an algorithm with variable spectral gain with dynamically adaptive strength
Kumar Comparative performance evaluation of MMSE-based speech enhancement techniques through simulation and real-time implementation
EP0998166A1 (en) Device for audio processing,receiver and method for filtering the wanted signal and reproducing it in presence of ambient noise
FR2894707A1 (en) METHOD FOR MEASURING THE PERCUSED QUALITY OF A DEGRADED AUDIO SIGNAL BY THE PRESENCE OF NOISE
Westhausen et al. Reduction of subjective listening effort for TV broadcast signals with recurrent neural networks
Tsilfidis et al. Blind single-channel suppression of late reverberation based on perceptual reverberation modeling
EP1792305A1 (en) Method and device for evaluating the efficiency of a noise reducing function for audio signals
Toyin et al. Speech enhancement in wireless communication system using hybrid spectral-kalman filter
EP1021805B1 (en) Method and apparatus for conditioning a digital speech signal
EP2515300A1 (en) Method and System for noise reduction
WO2020049263A1 (en) Device for speech enhancement by implementation of a neural network in the time domain
Pourmand et al. Computational auditory models in predicting noise reduction performance for wideband telephony applications
EP4287648A1 (en) Electronic device and associated processing method, acoustic apparatus and computer program
JP2003316380A (en) Noise reduction system for preprocessing speech- containing sound signal
CN114429763A (en) Real-time voice tone style conversion technology
Yu A Perceptual Model-Based Speech Enhancement Algorithm
Campbell et al. Single source noise reduction of received HF audio: experimental study
FR2845511A1 (en) System for improving the processing of signals, used in hearing aids, has arrangement which prevents parasitic signals

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20070831

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC NL PL PT RO SE SI SK TR

17Q First examination report despatched

Effective date: 20080310

DAX Request for extension of the european patent (deleted)
GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: FRANCE TELECOM

GRAS Grant fee paid

Free format text: ORIGINAL CODE: EPIDOSNIGR3

GRAA (expected) grant

Free format text: ORIGINAL CODE: 0009210

AK Designated contracting states

Kind code of ref document: B1

Designated state(s): AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LI LT LU LV MC NL PL PT RO SE SI SK TR

REG Reference to a national code

Ref country code: GB

Ref legal event code: FG4D

Free format text: NOT ENGLISH

REG Reference to a national code

Ref country code: CH

Ref legal event code: EP

REG Reference to a national code

Ref country code: IE

Ref legal event code: FG4D

REF Corresponds to:

Ref document number: 602006008111

Country of ref document: DE

Date of ref document: 20090910

Kind code of ref document: P

REG Reference to a national code

Ref country code: ES

Ref legal event code: FG2A

Ref document number: 2329932

Country of ref document: ES

Kind code of ref document: T3

NLV1 Nl: lapsed or annulled due to failure to fulfill the requirements of art. 29p and 29m of the patents act
PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: LT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: FI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: IS

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091129

Ref country code: AT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: SE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: PL

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: LV

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: NL

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: SI

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

REG Reference to a national code

Ref country code: IE

Ref legal event code: FD4D

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: PT

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091129

Ref country code: BG

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091029

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: CZ

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: IE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: EE

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: RO

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

Ref country code: DK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: SK

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

PLBE No opposition filed within time limit

Free format text: ORIGINAL CODE: 0009261

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: NO OPPOSITION FILED WITHIN TIME LIMIT

26N No opposition filed

Effective date: 20100503

BERE Be: lapsed

Owner name: FRANCE TELECOM

Effective date: 20100228

REG Reference to a national code

Ref country code: CH

Ref legal event code: PL

GBPC Gb: european patent ceased through non-payment of renewal fee

Effective date: 20100213

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: LI

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100228

Ref country code: MC

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100301

Ref country code: GR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20091030

Ref country code: CH

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100228

REG Reference to a national code

Ref country code: FR

Ref legal event code: ST

Effective date: 20101029

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: FR

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100301

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: BE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100228

Ref country code: DE

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100901

REG Reference to a national code

Ref country code: ES

Ref legal event code: FD2A

Effective date: 20110308

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: GB

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100213

Ref country code: IT

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100213

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: ES

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20110307

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: ES

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100214

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: CY

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: HU

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20100130

Ref country code: LU

Free format text: LAPSE BECAUSE OF NON-PAYMENT OF DUE FEES

Effective date: 20100213

PG25 Lapsed in a contracting state [announced via postgrant information from national office to epo]

Ref country code: TR

Free format text: LAPSE BECAUSE OF FAILURE TO SUBMIT A TRANSLATION OF THE DESCRIPTION OR TO PAY THE FEE WITHIN THE PRESCRIBED TIME-LIMIT

Effective date: 20090729