CN106297767A - Voice acquisition method based on speech recognition and system - Google Patents
Voice acquisition method based on speech recognition and system Download PDFInfo
- Publication number
- CN106297767A CN106297767A CN201610679482.7A CN201610679482A CN106297767A CN 106297767 A CN106297767 A CN 106297767A CN 201610679482 A CN201610679482 A CN 201610679482A CN 106297767 A CN106297767 A CN 106297767A
- Authority
- CN
- China
- Prior art keywords
- voice signal
- analog voice
- past
- enlargement ratio
- present day
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 76
- 230000003321 amplification Effects 0.000 claims abstract description 35
- 238000003199 nucleic acid amplification method Methods 0.000 claims abstract description 35
- 238000006243 chemical reaction Methods 0.000 claims abstract description 12
- 238000004088 simulation Methods 0.000 claims description 30
- 230000000694 effects Effects 0.000 claims description 18
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 230000008859 change Effects 0.000 abstract description 8
- 238000012545 processing Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000009931 harmful effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000002939 deleterious effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000032258 transport Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech 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
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (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)
- Signal Processing (AREA)
- Machine Translation (AREA)
- Circuit For Audible Band Transducer (AREA)
Abstract
The present invention provides a kind of voice acquisition method based on speech recognition and system.Wherein method includes: gather analog voice signal；According to the present day analog voice signal collected and in the past analog voice signal calculate enlargement ratio that present day analog voice signal is amplified, obtain current enlargement ratio；According to current enlargement ratio, present day analog voice signal is amplified, the present day analog voice signal being amplified；The present day analog voice signal of amplification is carried out analog digital conversion, obtains Contemporary Digital voice signal, and using Contemporary Digital voice signal as the input signal of speech recognition.Its current enlargement ratio being amplified present day analog voice signal is real-time change, a reasonable current enlargement ratio can be provided for present day analog voice signal, the present day analog voice signal after processing and amplifying is made not have distorted signals or the inadequate situation of precision, can be that speech recognition provides a good signal input basis, improve the discrimination of speech recognition.
Description
Technical field
The present invention relates to voice collecting field, particularly relate to a kind of voice acquisition method based on speech recognition and be
System.
Background technology
Present speech recognition is more ripe, and people also seek to improve phonetic recognization rate making great efforts untiringly
Method.One important step of speech recognition is exactly the collection of voice signal, and chip passes through mould the analogue signal collected
Number converter is converted into digital signal, as the input of speech recognition algorithm.Therefore the collection of voice signal is phonetic recognization rate
One important parameter of height, only obtains good digital signal and could provide a good base to speech recognition algorithm
Plinth, thus improve the discrimination of speech recognition.But traditional speech signal collection technology is providing signal for speech recognition algorithm
Generally the voice signal gathered is carried out the amplification of identical multiplying power during input, it is provided that signal input relatively rough, it is impossible to for language
Sound identification provides good digital signal input, and the discrimination causing speech recognition is the highest.
Summary of the invention
In consideration of it, be necessary to cause, for traditional voice Signal Collection Technology, the problem that the discrimination of speech recognition is the highest,
Voice acquisition method based on speech recognition and the system of a kind of discrimination that can improve speech recognition are provided.
For reaching goal of the invention, it is provided that a kind of voice acquisition method based on speech recognition, described method includes:
Gathering analog voice signal, wherein said analog voice signal includes present day analog voice signal and in the past simulates language
Tone signal；
According to the described present day analog voice signal collected and described in the past analog voice signal calculate to described currently
The enlargement ratio that analog voice signal is amplified, obtains current enlargement ratio；
According to described current enlargement ratio, described present day analog voice signal is amplified, the present day analog being amplified
Voice signal；
The present day analog voice signal of described amplification is carried out analog digital conversion, obtains Contemporary Digital voice signal, and by institute
State the Contemporary Digital voice signal input signal as speech recognition.
Wherein in an embodiment, present day analog voice signal that described basis collects and in the past analog voice signal
And preset algorithm acquires and includes with to the step of the current enlargement ratio that described present day analog voice signal is amplified:
Described present day analog voice letter is obtained according to described present day analog voice signal and described in the past analog voice signal
Number meansigma methods；
Obtain according to the maximum analog voice signal in described present day analog voice signal and described in the past analog voice signal
Take the outstanding value representing described present day analog voice signal optimum amplification effect；
Described current enlargement ratio is obtained according to described meansigma methods and described outstanding value.
Wherein in an embodiment, according to the present day analog voice signal collected and in the past analog voice signal acquisition
Obtain the step of the current enlargement ratio that described present day analog voice signal is amplified is included:
Obtain the in the past ideal of the in the past analog voice signal of the predetermined number neighbouring with described present day analog voice signal
Enlargement ratio；
According to described in predetermined number in the past preferable enlargement ratio and with described in each in the past preferable enlargement ratio corresponding
The enlargement ratio factor obtain described current enlargement ratio.
Wherein in an embodiment, the described acquisition predetermined number neighbouring with described present day analog voice signal is in the past
The step of the in the past preferable enlargement ratio of analog voice signal includes:
Obtain described in each historical simulation voice signal gathered before in the past analog voice signal, according to past each described
Time analog voice signal and historical simulation voice signal that described in each, in the past analogue signal is corresponding obtain described in each in the past
The in the past meansigma methods of analog voice signal；
The maximum history mould in historical simulation voice signal according in the past analog voice signal and correspondence described in each
Intend voice signal and obtain the most outstanding value representing in the past analog voice signal optimum amplification effect described in each；
Obtain described in each according to the in the past meansigma methods that described in each, in the past analog voice signal is corresponding and the most outstanding value
The in the past preferable enlargement ratio that in the past analog voice signal is corresponding.
Wherein in an embodiment, described in the past analog voice signal is the closer to described present day analog voice signal, institute
The proportion shared by the enlargement ratio factor stating the in the past preferable enlargement ratio of in the past analog voice signal corresponding is the biggest；
The described in the past enlargement ratio factor sum that preferable enlargement ratio is corresponding of predetermined number meets predetermined amount angle value.
The present invention also provides for a kind of speech collecting system based on speech recognition, and described system includes:
Acquisition module, is used for gathering analog voice signal, and wherein said analog voice signal includes that present day analog voice is believed
Number and in the past analog voice signal；
Acquisition module, the present day analog voice signal collected for basis and in the past analog voice signal calculate described
The enlargement ratio that present day analog voice signal is amplified, obtains current enlargement ratio；
Amplification module, for being amplified described present day analog voice signal according to described current enlargement ratio, obtains
The present day analog voice signal amplified；
Modular converter, for the present day analog voice signal of described amplification is carried out analog digital conversion, obtains Contemporary Digital language
Tone signal, and using described Contemporary Digital voice signal as the input signal of speech recognition.
Wherein in an embodiment, described acquisition module includes:
Meansigma methods acquiring unit, for obtaining according to described present day analog voice signal and described in the past analog voice signal
The meansigma methods of described present day analog voice signal；
Outstanding value acquiring unit, for according in described present day analog voice signal and described in the past analog voice signal
Maximum analog voice signal obtains the outstanding value representing described present day analog voice signal optimum amplification effect；
Enlargement ratio obtains unit, for obtaining described current enlargement ratio according to described meansigma methods and described outstanding value.
Wherein in an embodiment, described acquisition module includes:
First acquiring unit, for obtaining the in the past simulation language of the predetermined number neighbouring with described present day analog voice signal
The in the past preferable enlargement ratio of tone signal；
Second acquisition unit, for according to described in predetermined number in the past preferable enlargement ratio and with described in each in the past
The enlargement ratio factor that preferable enlargement ratio is corresponding obtains described current enlargement ratio.
Wherein in an embodiment, described first acquiring unit includes:
In the past meansigma methods obtains subelement, for obtaining described in each history mould gathered before in the past analog voice signal
Intend voice signal, according in the past analog voice signal and historical simulation that described in each, in the past analogue signal is corresponding described in each
Voice signal obtains the in the past meansigma methods of in the past analog voice signal described in each；
The most outstanding value obtains subelement, for the history mould according in the past analog voice signal and correspondence described in each
Intend the maximum historical simulation voice signal acquisition in voice signal and represent that described in each, in the past analog voice signal optimum amplifies effect
The most outstanding value of fruit；
In the past preferable enlargement ratio obtains subelement, according in the past corresponding average of analog voice signal described in each
Value and the most outstanding value obtain the in the past preferable enlargement ratio that described in each, in the past analog voice signal is corresponding.
Wherein in an embodiment, described in the past analog voice signal is the closer to described present day analog voice signal, institute
The proportion shared by the enlargement ratio factor stating the in the past preferable enlargement ratio of in the past analog voice signal corresponding is the biggest；
The described in the past enlargement ratio factor sum that preferable enlargement ratio is corresponding of predetermined number meets predetermined amount angle value.
The present invention also provides for a kind of speech collecting system based on speech recognition, and described system includes:
Speech signal collection device, is used for gathering analog voice signal, and wherein said analog voice signal includes present day analog
Voice signal and in the past analog voice signal；
Multiplying power arithmetical unit, for according to the present day analog voice signal collected and in the past analog voice signal calculating to institute
State the enlargement ratio that present day analog voice signal is amplified, obtain current enlargement ratio；
Analogue amplifier, is connected with described speech signal collection device and described multiplying power arithmetical unit, is used for receiving described voice
The described present day analog voice signal that signal picker gathers, and according to the described current times magnification of described multiplying power offer arithmetical unit
Described present day analog voice signal is amplified by rate, the present day analog voice signal being amplified；
Analog-digital converter, is connected with described analogue amplifier, applies also for being connected arithmetical unit with speech recognition, for institute
The present day analog voice signal stating amplification carries out analog digital conversion, obtains Contemporary Digital voice signal, and transports to described speech recognition
Calculate device and input described Contemporary Digital voice signal.
The beneficial effect comprise that
Above-mentioned voice acquisition method based on speech recognition and system, be amplified present day analog voice signal is current
Enlargement ratio is real-time change, current enlargement ratio or be the size according to present day analog voice signal and regulate in real time
Arrive, or be the in the past preferable enlargement ratio corresponding with in the past analog voice signal and be calculated, therefore can be present day analog
Voice signal provides a reasonable current enlargement ratio, makes the present day analog voice signal after processing and amplifying not have and puts
The most excessive and distorted signals or amplify the situation that too small precision is inadequate, and then can be that speech recognition provides a good signal
Input basis, improves the discrimination of speech recognition.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the voice acquisition method based on speech recognition in an embodiment；
Fig. 2 is the schematic flow sheet of the voice acquisition method based on speech recognition in another embodiment；
Fig. 3 is the schematic flow sheet of the voice acquisition method based on speech recognition in another embodiment；
Fig. 4 is the modular structure schematic diagram of the speech collecting system based on speech recognition in an embodiment；
Fig. 5 is the electrical block diagram of the speech collecting system based on speech recognition in an embodiment；
Fig. 6 is the schematic diagram of original input signal in an embodiment；
Fig. 7 is the schematic diagram of the output signal in an embodiment；
Fig. 8 is the signal of the output signal in another embodiment.
Detailed description of the invention
In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples pair
Present invention voice acquisition method based on speech recognition and system are further elaborated.Should be appreciated that described herein
Specific embodiment only in order to explain the present invention, be not intended to limit the present invention.
In one embodiment, as shown in Figure 1, it is provided that a kind of voice acquisition method based on speech recognition, the method
Comprise the following steps:
S100, gathers analog voice signal, and wherein analog voice signal includes present day analog voice signal and in the past simulates
Voice signal.
S200, according to the present day analog voice signal collected and in the past analog voice signal calculate to present day analog voice
The enlargement ratio that signal is amplified, obtains current enlargement ratio；
S300, is amplified present day analog voice signal according to current enlargement ratio, the present day analog language being amplified
Tone signal.
S400, carries out analog digital conversion by the present day analog voice signal of amplification, obtains Contemporary Digital voice signal, and ought
Front audio digital signals is as the input signal of speech recognition.
After voice acquisition method in the present embodiment collects analog voice signal, the analog voice signal that will collect
Storing, the analog voice signal collected so before is in the past analog voice signal, the analog voice currently collected
Signal is present day analog voice signal, when being amplified present day analog voice signal, no longer uses traditional to all simulations
Voice signal all uses identical enlargement ratio to be amplified, but according to present day analog voice signal and in the past analog voice letter
Number be calculated one can according to the size of present day analog voice signal the current enlargement ratio of Real-time and Dynamic change, Jin Ergen
Being amplified present day analog voice signal according to current enlargement ratio, the present day analog voice signal being amplified, then to putting
Big present day analog voice signal carries out analog digital conversion, obtains Contemporary Digital voice signal, is only changed by analog voice signal
For could be by speech recognition algorithm identification after audio digital signals, using defeated as speech recognition of this current audio digital signals
Enter signal, carry out speech recognition according to this input signal, it is possible to be effectively improved the discrimination of speech recognition.Due to it every time to working as
Front simulation voice signal be amplified process time use current enlargement ratio be all real-time change, current enlargement ratio or
It is the size according to present day analog voice signal and regulates in real time, or be with corresponding in the past preferable of in the past analog voice signal
Enlargement ratio and be calculated, it can provide a reasonable current enlargement ratio for present day analog voice signal, can preferably
Present day analog voice signal is amplified process, make the present day analog voice signal after processing and amplifying not have amplified
Big and distorted signals or amplify the situation that too small precision is inadequate, the present day analog voice signal after amplification can express its institute very well
Comprise information, after the present day analog voice signal after this amplification carries out analog digital conversion, be converted to the voice signal of numeral, this numeral
Voice signal, as the input signal of speech recognition, can be that speech recognition provides a good signal input basis, improve
The discrimination of speech recognition.
In one embodiment, seeing Fig. 2, step S200 includes:
S200a, according to present day analog voice signal and in the past analog voice signal obtain the flat of present day analog voice signal
Average.
S200b, according to present day analog voice signal and in the past the maximum analog voice signal in analog voice signal obtain
Represent the outstanding value of present day analog voice signal optimum amplification effect.
S200c, obtains current enlargement ratio according to meansigma methods and outstanding value.
For obtaining the one of the current enlargement ratio dynamically changed with the size of present day analog voice signal in the present embodiment
Individual detailed description of the invention.For speech recognition algorithm, if analogue signal is exaggerated excessive, then may distortion phenomenon,
Such as: speech recognition algorithm can be regarded as 1023 these digital signals the voltage more than 3.3V, and if analogue signal quilt
The multiple amplified is inadequate, then the digital signal be converted to also can be the least, causes precision the highest, thus provide one good current
Enlargement ratio is the basis obtaining good digital signal.There is provided a kind of in the present embodiment and can obtain good current amplification
The algorithm of multiplying power: the design outstanding value x of present day analog voice signal and meansigma methods y, meansigma methods be present day analog voice signal and
The meansigma methods of the in the past analog voice signal before gathered, it is preferred that meansigma methods is present day analog voice signal and gather before
The arithmetic mean of instantaneous value of in the past analog voice signal, outstanding value represents that present day analog voice signal reaches the number of optimum amplification effect
Value, it is generally recognized that when meansigma methods y of the present day analog voice signal gathered trends towards outstanding value x, it is believed that the voice letter collected
Number the most perfect.Preferably, in one embodiment, outstanding value be maximum historical simulation voice signal value 3/4ths.Logical
Often think when the meansigma methods of present day analog voice signal that collect and 3/4ths of the value of maximum historical simulation voice signal
Time close, the present day analog voice signal collected is best, can directly use, it is not necessary to be amplified again, is i.e. equivalent to amplify
Multiplying power is 1.For easy computing, generally using 3/4ths of the value of maximum historical simulation voice signal as outstanding value, this
Value can the amplification effect of reasonable reflection present day analog voice signal.Present day analog language is obtained according to meansigma methods and outstanding value
The current enlargement ratio z=z* (x/y) of tone signal, so when the present day analog voice signal collected is the biggest when, currently
The meansigma methods that analog voice signal is corresponding will be bigger than outstanding value, then the value of x/y is less than 1, and z* (x/y) is exactly to reduce amplification
Multiplying power, when the present day analog voice signal collected is the least when, the meansigma methods that present day analog voice signal is corresponding will compare
Outstanding value is little, then the value of x/y is greater than 1, and now z* (x/y) is exactly to increase enlargement ratio, thus reaches to be automatically adjusted
The effect of current enlargement ratio, thus provide a good signal input for speech recognition, improve the discrimination of speech recognition.
This embodiment has reasonable application in recording field, and the voice signal human ear using said method to deal sounds more
Good, it is possible to effectively to reduce the probability of sonic boom.
Seeing Fig. 6 and Fig. 7, Fig. 6 is the schematic diagram collecting present day analog voice signal in an embodiment, and Fig. 7 is for adopting
With the schematic diagram of the output signal that the method in the present embodiment obtains, it can be seen that use the method in the present embodiment
Can be good at amplifying original input signal.
In one embodiment, seeing Fig. 3, step S200 includes:
S210, obtains the in the past ideal of the in the past analog voice signal of the predetermined number neighbouring with present day analog voice signal
Enlargement ratio.
S220, according to the in the past preferable enlargement ratio of predetermined number and with each in the past preferable corresponding putting of enlargement ratio
The big multiplying power factor obtains current enlargement ratio.
The meansigma methods of the voice signal that above-mentioned basis collects and outstanding value is used to obtain the embodiment of current enlargement ratio
Although the suddenly big or suddenly small voice signal collected there to be preferable regulating effect, but owing to changing present day analog voice letter in real time
Number enlargement ratio data can be caused unnatural, speech recognition is had a lot of deleterious effect.Speech recognition is different from human ear,
Speech recognition is the discriminatory analysis to digital signal, and it all can be had an impact by the size of digital signal, the tone etc., so local
Zoom in or out sound and can affect the effect of speech recognition on the contrary.Therefore, the preset algorithm in the present embodiment is: use with current
The in the past preferable enlargement ratio of the in the past analog voice signal of the predetermined number that analog voice signal is neighbouring calculates current times magnification
Rate, the in the past analog voice signal thus according to the predetermined number before present day analog voice signal the current times magnification obtained
Rate, i.e. considers the in the past analog voice signal in the past a period of time, and it is the enlargement ratio carried out for the overall situation
Regulation rather than local zoom in or out acoustical signal, thus reach both can be automatically adjusted enlargement ratio, the most do not have office
The harmful effect that voice signal brings is amplified in portion so that more natural through amplifying the analog voice signal of output.See Fig. 6 and
Fig. 8, Fig. 6 are the schematic diagram collecting present day analog voice signal in an embodiment, and Fig. 8 is to use the side in the present embodiment
The schematic diagram of the output signal that method obtains, it can be seen that the output signal using the method in the present embodiment to obtain shows
The most smooth, the analog voice signal acquired is more natural.
Wherein, what deserves to be explained is, in the past analog voice signal is the closer to present day analog voice signal, in the past analog voice
The proportion shared by the enlargement ratio factor that the in the past preferable enlargement ratio of signal is corresponding is the biggest.The in the past preferable of predetermined number is amplified
Enlargement ratio factor sum corresponding to multiplying power meets predetermined amount angle value.I.e. the current enlargement ratio of present day analog voice signal be by
The in the past preferable enlargement ratio that above the in the past analog voice signal of predetermined number is corresponding determines, and the closer to present day analog language
The in the past preferable enlargement ratio that the in the past analog voice signal of tone signal is corresponding is the biggest on the impact of current enlargement ratio, the most both
Can reflect the overall situation impact on current enlargement ratio in a period of time, can obtain again preferably can be to present day analog voice signal
The current enlargement ratio being amplified, the analogue signal making output is more natural, improves good input for speech recognition further
Signal, thus improve the discrimination of speech recognition further.
Wherein it is desired to explanation, predetermined amount angle value represents that the in the past preferable enlargement ratio of predetermined number is on the whole to working as
The metric level of the impact of front enlargement ratio, this metric level can be arbitrary value, is obtaining currently putting of this metric level
After big multiplying power, do corresponding adjustment according to this current measurements rank, thus obtain and present day analog voice signal is put
Big current enlargement ratio.For easy computing, this predetermined amount angle value is preferably 1.
In one embodiment, step S210 includes:
S210a, obtains each historical simulation voice signal in the past gathered before analog voice signal, according to described in each
In the past analog voice signal and each historical simulation voice signal that in the past analogue signal is corresponding obtain each and in the past simulate language
The in the past meansigma methods of tone signal.
S210b, according to the maximum history in the historical simulation voice signal of each in the past analog voice signal and correspondence
Analog voice signal obtains the most outstanding value representing each in the past analog voice signal optimum amplification effect.
It is past that S210c, in the past analog voice signal is corresponding according to each in the past meansigma methods and the most outstanding value obtain each
Time in the past preferable enlargement ratio corresponding to analog voice signal.
This step be obtain each in the past preferable enlargement ratio that in the past analog voice signal is corresponding be embodied as step.
Each in the past preferable enlargement ratio can the dynamic Adjustment effect of each in the past analog voice signal of reasonable reflection, comprehensively
The impact on current enlargement ratio of multiple in the past preferable enlargement ratios, it is possible to reduce partial enlargement or reduce voice signal to language
The harmful effect of sound identification.
In one embodiment, predetermined number is 10.Quantity below in conjunction with in the past preferable enlargement ratio is 10
One specific embodiment is described in detail:
Obtain front 10 the in the past preferable amplifications that in the past analog voice signal corresponding neighbouring with present day analog voice signal
Multiplying power, respectively z1, z2, z3, z4, z5, z6, z7, z8, z9, z10, obtain each amplification that in the past preferable enlargement ratio is corresponding
The multiplying power factor, wherein, the enlargement ratio factor can be the most set in advance, and is saved in corresponding memory module
In, directly invoke during use, it is also possible to being dynamically change, it is the nearest that rule change meets distance present day analog voice signal
Enlargement ratio factor proportion corresponding in the past analog voice signal the biggest, and each enlargement ratio factor sum meets predetermined amount
The rule of angle value.According to 10 in the past preferable enlargement ratio level 10 the enlargement ratio factor that in the past preferable enlargement ratio is corresponding obtain
To current enlargement ratio.Preferably, in one embodiment, the current enlargement ratio z of present day analog voice signal is: z=a1*
z1+a2*z2+a3*z3+a4*z4+a5*z5+a6*z6+a7*z7+a8*z8+a9*z9+a10*z10；Wherein, a1 >=a2 >=a3 >=
a4≥a5≥a6≥a7≥a8≥a9≥a10；A1+a2+a3+a4+a5+a6+a7+a8+a9+a10=1.
Preferably, in a specific embodiment, a1=0.2, a2=0.18, a31=0.16, a4=0.14, a5=
0.10, a6=0.08, a71=0.06, a8=0.04, a91=0.02, a10=0.02.
What deserves to be explained is, the acquisition mode of each in the past preferable enlargement ratio obtains current amplification in previous embodiment
The mode of multiplying power is identical, is all to obtain by obtaining meansigma methods and outstanding value, and here is omitted.
In one embodiment, include after step S100: storage analog voice signal, when facilitating subsequent calculations, transfer mould
Intend voice signal.
In one embodiment, include after step S200: store current enlargement ratio, conveniently calculate next and currently amplify
Use as in the past preferable enlargement ratio during multiplying power.
In one embodiment, include after step S400: storage Contemporary Digital voice signal, conveniently carry out speech recognition
Call reading.
One of ordinary skill in the art will appreciate that all or part of flow process realizing in above-described embodiment method, be permissible
Instructing relevant hardware by computer program to complete, described program can be stored in a computer read/write memory medium
In, this program is upon execution, it may include such as the flow process of the embodiment of above-mentioned each method.Wherein, described storage medium can be magnetic
Dish, CD, read-only store-memory body (Read-Only Memory, ROM) or random store-memory body (Random Access
Memory, RAM) etc..
In one embodiment, as shown in Figure 4, additionally providing a kind of speech collecting system based on speech recognition, this is
System includes: acquisition module 100, is used for gathering analog voice signal, and wherein analog voice signal includes present day analog voice signal
In the past analog voice signal.Acquisition module 200, the present day analog voice signal collected for basis and in the past analog voice
Signal calculates the enlargement ratio being amplified present day analog voice signal, obtains current enlargement ratio.Amplification module 300, uses
According to current enlargement ratio, described present day analog voice signal being amplified, the present day analog voice signal being amplified.
Modular converter 400, for the present day analog voice signal of amplification is carried out analog digital conversion, obtains Contemporary Digital voice signal, and
Using Contemporary Digital voice signal as the input signal of speech recognition.
Present day analog voice signal is amplified the current times magnification processed by the speech collecting system in the present embodiment
Rate is real-time change, current enlargement ratio or be the size according to present day analog voice signal and regulate in real time and obtain, and
It is the in the past preferable enlargement ratio corresponding with in the past analog voice signal and is calculated, therefore can believe for present day analog voice
Number provide a reasonable current enlargement ratio, make the present day analog voice signal after processing and amplifying do not have amplification excessive
And distorted signals or amplify the situation that too small precision is inadequate, the present day analog voice signal after amplification can be expressed it very well and be wrapped
Containing information, and then can be that speech recognition provides a good signal input basis, improve the discrimination of speech recognition.
In one embodiment, acquisition module 200 includes: meansigma methods acquiring unit 200a, for according to present day analog language
Tone signal and in the past analog voice signal obtain the meansigma methods of described present day analog voice signal.Outstanding value acquiring unit 200b,
Work as obtaining expression according to the maximum analog voice signal in present day analog voice signal and described in the past analog voice signal
The outstanding value of front simulation voice signal optimum amplification effect.Enlargement ratio obtains unit 200c, for according to meansigma methods and outstanding
Value obtains current enlargement ratio.
In one embodiment, acquisition module 200 includes: the first acquiring unit 210, for obtaining and described present day analog
The in the past preferable enlargement ratio of the in the past analog voice signal of the predetermined number that voice signal is neighbouring.Second acquisition unit 220, uses
In the in the past preferable enlargement ratio according to predetermined number and with each enlargement ratio factor that in the past preferable enlargement ratio is corresponding
Obtain current enlargement ratio.
In one embodiment, the first acquiring unit 210 includes: in the past meansigma methods obtains subelement 210a, is used for obtaining
Each historical simulation voice signal in the past gathered before analog voice signal, according to each in the past analog voice signal and each
Historical simulation voice signal corresponding to individual in the past analogue signal obtains the in the past meansigma methods of each in the past analog voice signal.In the past
Outstanding value obtains subelement 210b, in the historical simulation voice signal according to each in the past analog voice signal and correspondence
Maximum historical simulation voice signal obtain the most outstanding value representing each in the past analog voice signal optimum amplification effect.Past
Time ideal enlargement ratio obtain subelement 210c, in the past analog voice signal is corresponding according to each in the past meansigma methods and the most excellent
Elegant value obtains each in the past preferable enlargement ratio that in the past analog voice signal is corresponding.
In one embodiment, in the past analog voice signal is the closer to present day analog voice signal, described in the past simulates language
The proportion shared by the enlargement ratio factor that the in the past preferable enlargement ratio of tone signal is corresponding is the biggest.The in the past ideal of predetermined number is put
The enlargement ratio factor sum that multiplying power is corresponding greatly meets predetermined amount angle value.
In one embodiment, also include: memory module 500, be used for storing analog voice signal, current enlargement ratio and
Contemporary Digital voice signal.
In one embodiment, the present invention also provides for a kind of speech collecting system based on speech recognition, and this system includes:
Speech signal collection device 10, is used for gathering analog voice signal, and wherein said analog voice signal includes that present day analog voice is believed
Number and in the past analog voice signal.Multiplying power arithmetical unit 20, for according to the present day analog voice signal collected and in the past simulating
Voice signal calculates the enlargement ratio being amplified described present day analog voice signal, obtains current enlargement ratio.Simulation is put
Big device 30, is connected with speech signal collection device and described multiplying power arithmetical unit, for receiving the current of speech signal collection device collection
Analog voice signal, and according to the current enlargement ratio of multiplying power offer arithmetical unit 20, present day analog voice signal is amplified,
The present day analog voice signal being amplified.Analog-digital converter 40, is connected with analogue amplifier, applies also for transporting with speech recognition
Calculate device 60 to connect, for the present day analog voice signal amplified is carried out analog digital conversion, obtain Contemporary Digital voice signal, and to
Speech recognition arithmetical unit 60 inputs Contemporary Digital voice signal.
The present embodiment is for realizing to provide the hardware of the speech collecting system of good signal input for speech recognition algorithm
Realizing device, it can provide the current enlargement ratio of a dynamic change for present day analog voice signal, after making processing and amplifying
Present day analog voice signal do not have the excessive and distorted signals of amplification or amplify the situation that too small precision is inadequate, it is possible to for language
Sound identification provides a good signal input basis, improves the discrimination of speech recognition.
Preferably, in one embodiment, speech signal collection device 10 is MIC (Microphone, mike) harvester,
There is preferable recording effect.
In one embodiment, also include: digital signal processor 50, connect arithmetical unit 20 with analog-digital converter 40 and multiplying power
Connect, apply also for being connected with speech recognition arithmetical unit 60, be used for storing analog voice signal, current enlargement ratio and Contemporary Digital
Voice signal, and the Contemporary Digital voice signal of storage is input to speech recognition arithmetical unit 60.
Owing to the principle of this system solution problem is similar, therefore to aforementioned a kind of voice acquisition method based on speech recognition
The enforcement of this system may refer to the enforcement of preceding method, repeats no more in place of repetition.
Each technical characteristic of embodiment described above can combine arbitrarily, for making description succinct, not to above-mentioned reality
The all possible combination of each technical characteristic executed in example is all described, but, as long as the combination of these technical characteristics is not deposited
In contradiction, all it is considered to be the scope that this specification is recorded.
Embodiment described above only have expressed the several embodiments of the present invention, and it describes more concrete and detailed, but also
Can not therefore be construed as limiting the scope of the patent.It should be pointed out that, come for those of ordinary skill in the art
Saying, without departing from the inventive concept of the premise, it is also possible to make some deformation and improvement, these broadly fall into the protection of the present invention
Scope.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (11)
1. a voice acquisition method based on speech recognition, it is characterised in that described method includes:
Gather analog voice signal, wherein said analog voice signal include present day analog voice signal and in the past analog voice believe
Number；
Calculate described present day analog according to the described present day analog voice signal collected and described in the past analog voice signal
The enlargement ratio that voice signal is amplified, obtains current enlargement ratio；
According to described current enlargement ratio, described present day analog voice signal is amplified, the present day analog voice being amplified
Signal；
The present day analog voice signal of described amplification is carried out analog digital conversion, obtains Contemporary Digital voice signal, and work as described
Front audio digital signals is as the input signal of speech recognition.
Voice acquisition method based on speech recognition the most according to claim 1, it is characterised in that described basis collects
Present day analog voice signal and in the past analog voice signal and preset algorithm acquire with to described present day analog voice
The step of the current enlargement ratio that signal is amplified includes:
Described present day analog voice signal is obtained according to described present day analog voice signal and described in the past analog voice signal
Meansigma methods；
Table is obtained according to the maximum analog voice signal in described present day analog voice signal and described in the past analog voice signal
Show the outstanding value of described present day analog voice signal optimum amplification effect；
Described current enlargement ratio is obtained according to described meansigma methods and described outstanding value.
Voice acquisition method based on speech recognition the most according to claim 1, it is characterised in that according to working as of collecting
Front simulation voice signal and in the past analog voice signal acquire be amplified described present day analog voice signal current
The step of enlargement ratio includes:
The in the past preferable of in the past analog voice signal obtaining the predetermined number neighbouring with described present day analog voice signal is amplified
Multiplying power；
According to described in predetermined number in the past preferable enlargement ratio and with in the past preferable corresponding the putting of enlargement ratio described in each
The big multiplying power factor obtains described current enlargement ratio.
Voice acquisition method based on speech recognition the most according to claim 3, it is characterised in that described acquisition is with described
The step of the in the past preferable enlargement ratio of the in the past analog voice signal of the predetermined number that present day analog voice signal is neighbouring includes:
Obtain described in each historical simulation voice signal gathered before in the past analog voice signal, according in the past mould described in each
Plan voice signal and the historical simulation voice signal that described in each, in the past analogue signal is corresponding obtain and in the past simulate described in each
The in the past meansigma methods of voice signal；
The maximum historical simulation language in historical simulation voice signal according in the past analog voice signal and correspondence described in each
Tone signal obtains the most outstanding value representing in the past analog voice signal optimum amplification effect described in each；
Obtain described in each in the past according to the in the past meansigma methods that described in each, in the past analog voice signal is corresponding and the most outstanding value
The in the past preferable enlargement ratio that analog voice signal is corresponding.
Voice acquisition method based on speech recognition the most according to claim 4, it is characterised in that described in the past simulate language
Tone signal is the closer to described present day analog voice signal, and the in the past preferable enlargement ratio of described in the past analog voice signal is corresponding
Proportion shared by the enlargement ratio factor is the biggest；
The described in the past enlargement ratio factor sum that preferable enlargement ratio is corresponding of predetermined number meets predetermined amount angle value.
6. a speech collecting system based on speech recognition, it is characterised in that described system includes:
Acquisition module, is used for gathering analog voice signal, wherein said analog voice signal include present day analog voice signal and
In the past analog voice signal；
Acquisition module, for according to the present day analog voice signal that collects and in the past analog voice signal calculating to described currently
The enlargement ratio that analog voice signal is amplified, obtains current enlargement ratio；
Amplification module, for being amplified described present day analog voice signal according to described current enlargement ratio, is amplified
Present day analog voice signal；
Modular converter, for the present day analog voice signal of described amplification is carried out analog digital conversion, obtains Contemporary Digital voice letter
Number, and using described Contemporary Digital voice signal as the input signal of speech recognition.
Speech collecting system based on speech recognition the most according to claim 6, it is characterised in that described acquisition module bag
Include:
Meansigma methods acquiring unit, for obtaining described according to described present day analog voice signal and described in the past analog voice signal
The meansigma methods of present day analog voice signal；
Outstanding value acquiring unit, for according to the maximum in described present day analog voice signal and described in the past analog voice signal
Analog voice signal obtains the outstanding value representing described present day analog voice signal optimum amplification effect；
Enlargement ratio obtains unit, for obtaining described current enlargement ratio according to described meansigma methods and described outstanding value.
Voice acquisition method based on speech recognition the most according to claim 6, it is characterised in that described acquisition module bag
Include:
First acquiring unit, for obtaining the in the past analog voice letter of the predetermined number neighbouring with described present day analog voice signal
Number in the past preferable enlargement ratio；
Second acquisition unit, for according in the past preferable enlargement ratio and in the past preferable with described in each described in predetermined number
The enlargement ratio factor that enlargement ratio is corresponding obtains described current enlargement ratio.
Speech collecting system based on speech recognition the most according to claim 8, it is characterised in that described first obtains list
Unit includes:
In the past meansigma methods obtains subelement, for obtaining described in each historical simulation language gathered before in the past analog voice signal
Tone signal, according in the past analog voice signal and historical simulation voice that described in each, in the past analogue signal is corresponding described in each
The in the past meansigma methods of signal acquisition in the past analog voice signal described in each；
The most outstanding value obtains subelement, for the historical simulation language according in the past analog voice signal and correspondence described in each
Maximum historical simulation voice signal in tone signal obtains and represents described in each in the past analog voice signal optimum amplification effect
The most outstanding value；
In the past preferable enlargement ratio obtains subelement, according to the in the past meansigma methods that described in each, in the past analog voice signal is corresponding and
The most outstanding value obtains the in the past preferable enlargement ratio that described in each, in the past analog voice signal is corresponding.
Speech collecting system based on speech recognition the most according to claim 9, it is characterised in that described in the past simulate
Voice signal is the closer to described present day analog voice signal, and the in the past preferable enlargement ratio of described in the past analog voice signal is corresponding
The proportion shared by the enlargement ratio factor the biggest；
The described in the past enlargement ratio factor sum that preferable enlargement ratio is corresponding of predetermined number meets predetermined amount angle value.
11. 1 kinds of speech collecting systems based on speech recognition, it is characterised in that described system includes:
Speech signal collection device, is used for gathering analog voice signal, and wherein said analog voice signal includes present day analog voice
Signal and in the past analog voice signal；
Multiplying power arithmetical unit, for according to the present day analog voice signal collected and in the past analog voice signal calculate to described ought
The enlargement ratio that front simulation voice signal is amplified, obtains current enlargement ratio；
Analogue amplifier, is connected with described speech signal collection device and described multiplying power arithmetical unit, is used for receiving described voice signal
The described present day analog voice signal that harvester gathers, and according to the described current enlargement ratio pair of described multiplying power offer arithmetical unit
Described present day analog voice signal is amplified, the present day analog voice signal being amplified；
Analog-digital converter, is connected with described analogue amplifier, applies also for being connected arithmetical unit with speech recognition, for putting described
Big present day analog voice signal carries out analog digital conversion, obtains Contemporary Digital voice signal, and to described speech recognition arithmetical unit
Input described Contemporary Digital voice signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610679482.7A CN106297767B (en) | 2016-08-16 | 2016-08-16 | Voice acquisition method and system based on speech recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610679482.7A CN106297767B (en) | 2016-08-16 | 2016-08-16 | Voice acquisition method and system based on speech recognition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106297767A true CN106297767A (en) | 2017-01-04 |
| CN106297767B CN106297767B (en) | 2019-11-12 |
Family
ID=57679505
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610679482.7A Active CN106297767B (en) | 2016-08-16 | 2016-08-16 | Voice acquisition method and system based on speech recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106297767B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH04367899A (en) * | 1991-06-14 | 1992-12-21 | Ricoh Co Ltd | Agc control system of voice recognition device |
| JPH11194797A (en) * | 1997-12-26 | 1999-07-21 | Kyocera Corp | Speech recognition operating device |
| CN1700603A (en) * | 2004-12-31 | 2005-11-23 | 北京中星微电子有限公司 | Apparatus and method for digitalizing analog signal |
| CN101004673A (en) * | 2005-09-20 | 2007-07-25 | 三星电子株式会社 | Apparatus to convert analog signal of array microphone into digital signal and computer system including the same |
| CN101315770A (en) * | 2008-05-27 | 2008-12-03 | 北京承芯卓越科技有限公司 | System on speech recognition piece and voice recognition method using the same |
| CN101454973A (en) * | 2006-05-30 | 2009-06-10 | 冲电气工业株式会社 | Automatic gain controller |
-
2016
- 2016-08-16 CN CN201610679482.7A patent/CN106297767B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH04367899A (en) * | 1991-06-14 | 1992-12-21 | Ricoh Co Ltd | Agc control system of voice recognition device |
| JPH11194797A (en) * | 1997-12-26 | 1999-07-21 | Kyocera Corp | Speech recognition operating device |
| CN1700603A (en) * | 2004-12-31 | 2005-11-23 | 北京中星微电子有限公司 | Apparatus and method for digitalizing analog signal |
| CN101004673A (en) * | 2005-09-20 | 2007-07-25 | 三星电子株式会社 | Apparatus to convert analog signal of array microphone into digital signal and computer system including the same |
| CN101454973A (en) * | 2006-05-30 | 2009-06-10 | 冲电气工业株式会社 | Automatic gain controller |
| CN101315770A (en) * | 2008-05-27 | 2008-12-03 | 北京承芯卓越科技有限公司 | System on speech recognition piece and voice recognition method using the same |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106297767B (en) | 2019-11-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109712626B (en) | Voice data processing method and device | |
| CN102007776B (en) | Auditory prosthesis | |
| CN104144374B (en) | Assisting hearing method and system based on mobile device | |
| CN104980337B (en) | A kind of performance improvement method and device of audio processing | |
| CN106782584A (en) | Audio signal processing apparatus, method and electronic equipment | |
| CN109121057A (en) | A kind of method and its system of intelligence hearing aid | |
| CN107919133A (en) | For the speech-enhancement system and sound enhancement method of destination object | |
| CN106648527A (en) | Volume control method, device and playing equipment | |
| Wright et al. | Perceptual loss function for neural modeling of audio systems | |
| CN107734126A (en) | voice adjusting method, device, terminal and storage medium | |
| JPWO2006011405A1 (en) | Digital filtering method, digital filter device, digital filter program, computer-readable recording medium, and recorded device | |
| CN102164203A (en) | Information processing device and method and program | |
| CN105845149B (en) | The high acquisition methods of keynote and system in voice signal | |
| CN108235181A (en) | The method of noise reduction in apparatus for processing audio | |
| CN104936651B (en) | To for making cochlea implantation system adapt to the system and method that the customization acoustics scene of patient is rendered | |
| CN108877831A (en) | Blind source separating fast method and system based on multi-standard fusion frequency point screening | |
| CN110060696A (en) | Sound mixing method and device, terminal and readable storage medium storing program for executing | |
| JP2012083746A (en) | Sound processing device | |
| CN107369441A (en) | Noise-eliminating method, device and the terminal of voice signal | |
| CN104219390A (en) | Mobile terminal and sound recording method and device thereof | |
| CN103168479B (en) | Anti-singing device, sonifer, singing suppressing method and integrated circuit | |
| CN106297767A (en) | Voice acquisition method based on speech recognition and system | |
| CN103096230A (en) | All-digital type hearing-aid and changing channel matching and compensating method thereof | |
| CN112235679B (en) | Signal equalization method and processor suitable for earphone and earphone | |
| CN111491245B (en) | Digital hearing aid sound field identification algorithm based on cyclic neural network and implementation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| CB02 | Change of applicant information |
Address after: 519085 Guangdong city of Zhuhai province Jida West Road No. 107 Building 9 Building (1-4) Applicant after: Zhuhai jelee Polytron Technologies Inc Address before: 519085 Guangdong city of Zhuhai province Jida West Road No. 107 Building 9 Building Applicant before: Zhuhai Jieli Technology Co., Ltd. |
|
| COR | Change of bibliographic data | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder |
Address after: 519000 No. 333, Kexing Road, Xiangzhou District, Zhuhai City, Guangdong Province Patentee after: ZHUHAI JIELI TECHNOLOGY Co.,Ltd. Address before: Floor 1-107, building 904, ShiJiHua Road, Zhuhai City, Guangdong Province Patentee before: ZHUHAI JIELI TECHNOLOGY Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |