CN101459934B - Voice quality loss estimation method and related apparatus - Google Patents
Voice quality loss estimation method and related apparatus Download PDFInfo
- Publication number
- CN101459934B CN101459934B CN2007101724087A CN200710172408A CN101459934B CN 101459934 B CN101459934 B CN 101459934B CN 2007101724087 A CN2007101724087 A CN 2007101724087A CN 200710172408 A CN200710172408 A CN 200710172408A CN 101459934 B CN101459934 B CN 101459934B
- Authority
- CN
- China
- Prior art keywords
- voice quality
- radio link
- quality loss
- link parameter
- estimation
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000011156 evaluation Methods 0.000 claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims description 31
- 238000006243 chemical reaction Methods 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 8
- 230000001537 neural effect Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 7
- 238000012417 linear regression Methods 0.000 claims description 6
- 230000005540 biological transmission Effects 0.000 abstract description 6
- 230000008859 change Effects 0.000 abstract description 5
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 7
- 238000012549 training Methods 0.000 description 7
- 238000012546 transfer Methods 0.000 description 6
- 238000010276 construction Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 230000007423 decrease Effects 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241000239290 Araneae Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
Images
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a voice quality loss evaluation method and a correlative device, wherein the method comprises obtaining a wireless chain circuit parameter in wireless signal, and evaluating the voice quality loss according to the wireless chain circuit parameter and a preset parameter for evaluating voice quality loss. Compared with the voice quality forecasted and received by the prior art, relative change of voice transferred in wireless network is forecasted through employing the wireless chain circuit parameter, which can obtain better correlation, forecast more precisely, and is beneficial to rightly evaluating voice transmission in wireless network.
Description
Technical field
The present invention relates to communication technical field, be specifically related to voice quality loss estimation method and relevant apparatus.
Background technology
In modern wireless communication networks, speech business occupies consequence.In order to provide good voice quality service to the user, operator need do one to the voice quality of network and detect accurately.Voice quality can obtain by some objective evaluation instruments end to end, and this operation flow includes spider lines and 2 parts of wireless network, and wherein wireless network is the key link that influences voice quality.But can only measure a series of radio link parameters in wireless network, comprise the error rate (BER), delete frame per second (FER), switching frequency and switching state etc., they are with not directly contact of end-to-end speech quality itself.
Use these radio link parameters so, set up a mapping mode through combination with the end-to-end speech quality, promptly, can probably assess the impression of receiving terminal user's voice quality, and then learn the quality of voice transmission of wireless system from radio link quality by this mapping mode.
Existing a kind of technical scheme is:
At first obtain the radio link parameter of system, comprising: BER, FER, switching frequency and switching state etc.
By statistics to above-mentioned parameter, consider different phonetic vocoder characteristic is done corresponding adjustment, can draw the mark of voice quality indication (Speech Quality Indicator) scoring through computing independently.
Detailed process comprises: at first be to receive numerous radio link parameters, such as BER, FER, incoming level (RxLev) switching state etc., these parameters comprise wireless time domain specification, comprise fading rate, decline length, dafing depth, signal to noise ratio, signal interference ratio, signal level, handoff scenario etc.These parameters calculate the time domain available information again, these information can be used for calculating a series of time domain parameters that can be used for statistical analysis, such as any predefine (0.1s～maximum 10s), minimum value, mean value, standard variance, autocorrelation value in the period.Time domain parameter merge to produce a series of more near the relevant parameter of voice quality.Last fallout predictor obtains the predicted value of voice quality with these relevant parameters.Fallout predictor can be made up of neural net based on linearity or nonlinear prediction, or the dynamic change factor of configuration state machine response to network is such as change between mobile station speed, the non-frequency modulation of frequency modulation etc.
A simple prediction formula is Estimate=A (Parameter1)+B (Parameter2)+, and wherein coefficient A, B etc. can adopt the training sequence linear regression to obtain, and also can adopt nonlinear prediction to be worth more accurately.
Prediction formula above corresponding, some use more link parameter at present substitution, and formula that must we were familiar with up till now is
。For GSM full-speed voice encryption algorithm (GSM-FR), under the situation of not considering to switch with DTX, a formula that can provide is:
SQI＝20.67-57.2×BER-29.3×TFER-0.11×LFE
Wherein, calculation interval is 2.5s, and BER is the average error rate in the 2.5s in the past, and FER is the average frame per second of deleting in the 2.5s in the past, and TFER is meant the square root of FER.The maximum of restricted T FER is 0.66, and LFE is the longest frame length of deleting in the past 2.5s.Similar with RXQUAL, SQI is that every 0.5s calculates once.Also can change to 0.48s calculates once.
Can learn in above-mentioned the description of prior art scheme, after having determined to need to participate in the prediction radio link parameter, under the situation of the value of given voice quality and radio link parameter, coefficients such as A, B, C wherein, can adopt the training sequence linear regression to obtain, can certainly adopt nonlinear prediction to be worth more accurately.
The inventor simulates above-mentioned existing technical scheme, as shown in Figure 1, be to adopt 20 speech samples, radio link parameter is an example to delete frame condition, obtain the lossless of 20 speech samples and after each deletes all frames of traversal under the frame condition, all frames are perception estimation (the Perceptual Evaluation of Speech Quality of the voice quality that obtains after average through overtesting, PESQ) MOS score value, wherein PESQ is present widely used objective voice quality evaluating method.
Abscissa among Fig. 1 is 20 voice sequences, and ordinate is the MOS score value of voice quality.The same point of deleting frame condition is linked to be a broken line.Broken line 0 to broken line 13 represents that respectively 20 voice sequences are at the different speech assessments of deleting under the frame condition.
Fig. 1 middle polyline label is as shown in table 1 with the contrast tabulation of deleting frame condition:
Table 1
| Numbering | |
| 0? | Lossless |
| 1? | Lose 1 |
| 2? | 2 frames are lost 3 frames at |
| 3? | 2 frames are lost 2 frames at |
| 4? | Lose 2 frames continuously |
| 5? | 1 frame is lost 2 frames at |
| 6? | Lose 3 frames continuously |
| 7? | 1 frame is lost 3 frames at |
| 8? | 2 frames are lost 4 frames at |
| 9? | Lose 4 frames continuously |
| 10? | 1 frame is lost 4 frames at |
| 11? | 2 frames are lost 10 frames at |
| 12? | Lose 10 frames continuously |
| 13? | 1 frame is lost 10 frames at interval |
The data that provide according to Fig. 1 adopt the mode of above-mentioned SQI to carry out linear regression fit below, because just simulate the situation of frame losing among the figure, do not have simulated bit error, so the parameter in the formula is FER and LFE, do not have BER.The linear fit function LINEST that the electrical form software Excel of employing Microsoft company provides fits, and it is as follows to obtain the result:
Formula after the match is:
Coefficient of determination=0.34556
The standard variance of residual sum of squares (RSS)=0.128787.
Coefficient of determination is the ratio of estimated value with the desirable actual value of functional value, and scope if be 1 then sample has good correlation, does not have difference between the estimated value of Y and the actual value between 0～1.If coefficient of determination is 0, then regression formula can not be used for the anticipation function value.
In research and practice process to prior art, the inventor finds that there is following problem in prior art:
The coefficient of determination that obtains through above-mentioned analysis is 0.34556, the standard variance of residual sum of squares (RSS) is 0.128787, as seen SQI is not strong with the correlation that receives between the voice quality, moreover this is the several simulations of deleting frame mean value under clear speech samples, and the correlation under the actual conditions may be poorer.It is bigger to predict that in this way the voice quality result departs from actual conditions, and difficulty is made right judgement to the voice transfer situation of wireless network.
Summary of the invention
The technical problem that the embodiment of the invention solves provides voice quality loss estimation method and relevant apparatus, can obtain better correlation, and is more accurate to the voice transfer situation prediction of wireless network.
The embodiment of the invention provides a kind of evaluation method of voice quality loss, comprising:
Obtain the radio link parameter in the wireless signal, described radio link parameter comprises at least: delete frame per second;
Parametric coefficients estimation voice quality loss according to described radio link parameter and the estimation voice quality loss of presetting.
The embodiment of the invention provides a kind of device of voice quality loss estimation, comprising:
The radio link parameter extraction unit is used for obtaining the radio link parameter of wireless signal, and described radio link parameter comprises at least: delete frame per second;
Evaluation unit is used for the parametric coefficients estimation voice quality loss of radio link parameter that obtains according to described radio link parameter extraction unit and the estimation voice quality loss of presetting.
The embodiment of the invention provides a kind of device of estimating the parametric coefficients of voice quality loss, comprising:
The radio link parameter acquiring unit is used to obtain the radio link parameter for the treatment of examining system, and described radio link parameter comprises at least: delete frame per second;
Voice quality loss acquiring unit is used for calculating according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal the voice quality loss value of radio link parameter correspondence;
Coefficient calculation unit is used for carrying out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter.
Adopt technique scheme, embodiment of the invention beneficial technical effects is:
In the embodiment of the invention, obtain the radio link parameter in the wireless signal; And according to the parametric coefficients estimation voice quality loss of the described radio link parameter and the estimation voice quality loss of presetting.Relative variation by radio link parameter prediction wireless network transmissions voice, with respect to predicting the voice quality itself that receives in the prior art, can obtain better correlation, it is more accurate to predict, helps the voice transfer situation of wireless network is made correct evaluation.
Description of drawings
The voice quality scoring schematic diagram of Fig. 1 for the prior art scheme is simulated;
Fig. 2 is the flow chart of the embodiment of the invention one voice quality loss estimation method;
Fig. 3 is the flow chart of approximating method of the parametric coefficients of the embodiment of the invention two estimation voice quality losses;
Fig. 4 is the quality score difference schematic diagram of 20 sequence voice under the different radio link parameter in the embodiment of the invention;
Fig. 5 carries out the system construction drawing of linear fit to the coefficient of estimation voice quality loss for the embodiment of the invention;
Fig. 6 is the system construction drawing of the practical application mode of voice-over-net loss in the embodiment of the invention;
Fig. 7 is the structural representation of the device of the embodiment of the invention three voice quality loss estimations;
Fig. 8 is the structural representation of the device of the embodiment of the invention four voice quality loss estimations;
Fig. 9 is the structural representation of device of the parametric coefficients of the embodiment of the invention five estimation voice quality losses;
Figure 10 is the structural representation of device of the parametric coefficients of the embodiment of the invention six estimation voice quality losses.
Embodiment
The embodiment of the invention provides voice quality loss estimation method and relevant apparatus, can obtain better correlation, and is more accurate to the voice transfer situation prediction of wireless network.
Can learn that from existing technical scheme being used as training sequence with the receiving terminal voice quality, to fit the SQI correlation not unsatisfactory, we need set up a new effective mapping mode.As can see from Figure 1, the receiving terminal voice quality descends to some extent compared with the transmitting terminal nondestructive voice quality, but delete curve shape under the frame condition all near the curve of lossless case for every kind, promptly deleting under the frame condition, the decline of voice quality is near a constant, and this constant is with deleting frame per second and to delete frame continuously relevant.As seen, adopt the drop-out value just voice quality loss of receiving terminal, just can obtain compared with the higher correlation of top simple employing receiving terminal voice quality to the transmitting terminal voice quality.
Below voice quality loss estimation method provided by the invention and relevant apparatus are described in detail.
Embodiment one, and a kind of voice quality loss estimation method, flow chart comprise as shown in Figure 2:
A1 obtains the radio link parameter in the wireless signal;
In the embodiment of the invention, described radio link parameter comprises at least: the error rate, delete frame per second, the longest wherein of deleting frame length, received signal level, soft information, speech energy parameter or handover statistic continuously.
A2 is according to the parametric coefficients estimation voice quality loss of described radio link parameter and the estimation voice quality loss of presetting.
In the embodiment of the invention, estimate that according to the parametric coefficients of described radio link parameter and the estimation voice quality loss of presetting the voice quality loss can comprise:
At first, described radio link parameter is transformed into the form relevant with the voice quality loss;
In the embodiment of the invention, described variation can comprise:
By any one computing or the operation carried out in logarithm, index, power, evolution, quadrature, exponentiation or the weighted average windowing radio link parameter is carried out formal argument.
Be understandable that variation can also be other multiple mathematic(al) manipulation modes, the purpose of conversion is in order to obtain the form relevant with the voice quality loss, and helps the estimation of back.Concrete variation is not construed as limiting the invention.
Then, according to of the parametric coefficients estimation voice quality loss of described radio link parameter through form after the conversion and the estimation voice quality loss of presetting.
Be understandable that, described radio link parameter also can be without formal argument, directly utilize the radio link parameter estimation voice quality loss of obtaining, but can have better correlation through formal argument and voice quality loss for some radio link parameters.The parametric coefficients of described estimation voice quality loss can obtain concrete estimation mode reference example two by the algorithm estimation of presetting.
In the embodiment of the invention, the process of described estimation voice quality loss can adopt linear estimation device to carry out estimation or adopt non-linear estimation device to carry out estimation, can also utilize neural net to carry out estimation.
In the embodiment of the invention one, obtain the radio link parameter in the wireless signal; And according to the parametric coefficients estimation voice quality loss of the described radio link parameter and the estimation voice quality loss of presetting.Relative variation by radio link parameter prediction wireless network transmissions voice quality, with respect to predicting the voice quality itself that receives in the prior art, can obtain better correlation, it is more accurate to predict, helps the voice transfer situation of wireless network is made correct evaluation.
One of ordinary skill in the art will appreciate that all or part of step that realizes in the foregoing description method is to instruct relevant hardware to finish by program, described program can be stored in a kind of computer-readable recording medium, this program comprises the steps: when carrying out
Obtain the radio link parameter in the wireless signal;
Parametric coefficients estimation voice quality loss according to described radio link parameter and the estimation voice quality loss of presetting.
The above-mentioned storage medium of mentioning can be a read-only memory, disk or CD etc.
In the foregoing description one voice quality loss estimation method, the embodiment of the invention also provides a kind of approximating method of estimating the parametric coefficients of voice quality loss, can draw the present invention thus than better correlation of prior art scheme and fitting precision.
Embodiment two, a kind of approximating method of estimating the parametric coefficients of voice quality loss, and flow chart comprises as shown in Figure 3:
B1 obtains the radio link parameter for the treatment of examining system;
B2 obtains the voice quality loss value of described radio link parameter correspondence;
The described voice quality loss value for the treatment of examining system that obtains can comprise:
Calculate the voice quality loss value of radio link parameter correspondence according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal.
B3 carries out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter.
In the embodiment of the invention,, equally also be the form relevant that earlier described radio link parameter is transformed into the voice quality loss in order to obtain better correlation;
Carry out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter through the form after the conversion again.
In the embodiment of the invention, can be by the training sequence of the radio link parameter behind the variation and its corresponding voice quality loss value, carry out repeatedly the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss.
Described The Fitting Calculation is that linear regression fit calculates or nonlinear prediction The Fitting Calculation or neural network calculating.
One of ordinary skill in the art will appreciate that all or part of step that realizes in the foregoing description method is to instruct relevant hardware to finish by program, described program can be stored in a kind of computer-readable recording medium, this program comprises the steps: when carrying out
Obtain the radio link parameter for the treatment of examining system;
Obtain the voice quality loss value of described radio link parameter correspondence;
Carry out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter.
The above-mentioned storage medium of mentioning can be a read-only memory, disk or CD etc.
Method to the foregoing description two is specifically described below.
Consult Fig. 4, be in the embodiment of the invention in the quality score difference schematic diagram of 20 sequence voice under the different radio link parameter.Radio link parameter employing among this figure is deleted frame condition and is simulated.
Abscissa is represented 20 voice sequences among the figure, and ordinate is represented the difference (transmitting terminal speech assessment-receiving terminal speech assessment) of voice quality scoring, from the bottom to top the different frame conditions of deleting in 13 curve representation differences 13.Identical frame loss condition is connected to a curve, wherein, losing 1 frame, to represent to travel through the voice quality damage of losing behind 1 frame that obtains behind all frames average, with sequence 1 is example, it has 400 frames, lose 1 frame average speech mass loss=(lose the voice quality loss that the voice quality loss that obtains behind the 1st frame+lose obtains behind the 2nd frame+... lose the voice quality loss that the 400th frame obtains)/400, the similar calculating of all the other each sequences.Lose 2 frames continuously and represent to lose 2 frames, this 2 frame is a continuous position, such as 1/2, or 2/3 etc., equally also obtain an average speech mass loss after compiling all frames, be example with sequence 1, lose 2 frame average speech mass loss=(lose the 1st continuously, the voice quality loss that the voice quality loss that obtains behind 2 frames+lose obtains behind the 2nd, 3 frame+... lose the voice quality loss that the 399th, 400 frame obtains)/399.All the other continuous frame losings are similar.Frame losing represents that several frames of losing simultaneously are at interval at interval, such as 1,3, or 2,4.And 2 frames represent that several frames of losing simultaneously are interval 2 frames at interval, such as 1,4, or 2,5.
Fig. 4 middle polyline label is as shown in table 2 with the contrast tabulation of deleting frame condition:
Table 2
| Label | |
| 1? | Lose 1 |
| 2? | 2 frames are lost 2 frames at |
| 3? | Lose 2 frames continuously |
| 4? | 1 frame is lost 2 frames at |
| 5? | 2 frames are lost 3 frames at |
| 6? | Lose 3 frames continuously |
| 7? | 1 frame is lost 3 frames at |
| 8? | 2 frames are lost 4 frames at |
| 9? | Lose 4 frames continuously |
| 10? | 1 frame is lost 4 frames at |
| 11? | 2 frames are lost 10 frames at |
| 12? | Lose 10 frames continuously |
| 13? | 1 frame is lost 10 frames at interval |
[0090]Parameter in the present embodiment is: square root (TFER) and the longest frame length (LFE) of deleting of deleting frame per second.
Because of the decline that transmission causes voice quality to be marked, can see that the value of the data point in this time is much smaller compared with Fig. 1.
Below 20 above-mentioned voice sequences are carried out linear fit and calculate, The Fitting Calculation adopts the LINEST linear fit of Excel, result such as following table 3.
Table 3
| 8.2E-156? | 1.969201? | 260? |
| 1.681408? | 41.20929? | ? |
| 0.002216? | 0.083765? | 0? |
| 0.001318? | 0.002033? | #N/A? |
| 0.937227? | 0.047352? | #N/A? |
| 1926.026? | 258? | #N/A? |
| 8.637279? | 0.578501? | #N/A? |
| 0.182616? | 0.047261? | ? |
The formula of output is this time of Δ SQ=MOSx-MOSr=0.083765 * TFER+0.002216 * LFE, judgement factor reaches 0.937227, has approached very much 1.And the residual sum of squares (RSS) standard variance has only 0.047261.The numerical value spy that simulation obtains compared with existing technical scheme also has been much smaller.
As seen the correlation of the loss value of voice quality and radio link parameter is compared with the correlation of radio link parameter with receiving terminal voice quality itself, and is better.
Be reflected on the formula, that is:
ΔSQ＝MOSx-MOSr＝x×BER+y×TFER+z×LFE
Note not having constant term in the formula.
The measurement of MOSx and MOSr can obtain by subjective testing in the top formula, but is to adopt various objective voice survey tools to obtain more, uses very wide PESQ such as present, or is applied to the E-MODEL of VOIP network.Use PESQ and calculate MOSx and MOSr, thereby obtain Δ SQ, again radio link parameters such as BER, FER, LFE are carried out linear regression and can obtain wireless speech mass loss comparatively accurately.
The present invention in actual applications, because there is bigger correlation in the wireless network vocal quality loss with radio link parameter, but the concrete mapping mode of this correlation is relevant with the real network situation, and the variation of each parameter and coefficient need obtain by the training sequence prediction.
See also Fig. 5 carries out linear fit to the coefficient of estimation voice quality loss for the embodiment of the invention system construction drawing; This figure has embodied training process.
At first provide a database that comprises the received pronunciation sequence among Fig. 5, in theory in the various voice that the voice-over-net service call occurs preferably are included in, the result who fits out so just has one " completeness ", but in fact such voice are too many, database volume is limited, and the probability that various voice occur does not wait, and satisfies the voice sequence of using under most of scene as long as we choose, and such sequence can be not very little yet.On general, the sequence of choosing all is the clear speech samples under the no environment noise, and it will contain male voice female voice, different language, all ages and classes section etc.In addition, the voice activation ratio also is the factor that needs are paid close attention to, and generally needs the activation ratio between 40%～80%.Can obtain the MOS score value of these sequences, i.e. MOSx by scoring software.
The received pronunciation sequence is poured in the network system to be measured, and voice quality MOSr and a series of radio link parameter after receiving terminal measures loss comprise BER, FER, LFE, RxLevel, DTX ratio or the like.Calculate Δ MOS=MOSx-MOSr, these radio link parameters are carried out the time domain formal argument, carry out according to principles such as least mean-square errors afterwards that linearity fits or multinomial fits, just can obtain comprising the formula that fits of wireless parameter version and coefficient.
The time domain version of wireless parameter comprises mean value, maximum, variance, and their each exponential form etc., in order to obtain the fitted results of a best meaning, may need test of many times.Present available preliminary conclusion is that the voice loss is linear to FER 0.5 time (being TFER), and more approaching exponential number is 0.6667 time.
Fit the result who obtains above the employing and be applied in this real network, promptly can be as tolerance to voice-over-net mass loss degree.
See also Fig. 6 in the embodiment of the invention, the system construction drawing of the practical application mode of voice-over-net loss.
Among Fig. 6, the transmitting terminal of real network sends voice, behind wireless environment decay error code, by the receiving terminal received speech signal, simultaneously, measure the radio link parameter of this period, train the formula that fits that obtains above adopting, calculate the wireless network voice loss of this moment usefulness.The loss that calculates is big more, and this moment, the network state of correspondence was poor more.
TX among Fig. 6 is portable terminal (as: mobile phone), and RX is the base station, perhaps conversely.
The period of statistics is according to the measurement capability of receiving device to the network link parameter, in general short more, detection to network quality is accurate more, but neither be short more good more, too many measurement task can increase the weight of the load of system, and the training sequence that adopts is also wanted corresponding shortening, and too short sequence can have influence on the accuracy of voice measuring instrument.In the GSM network, network generally reports with the 480ms superframe period, and the multiple that can adopt 480ms this moment generally is advisable with 2.4s as measuring duration.
The device of enforcement the foregoing description method that the embodiment of the invention is provided describes below.
Embodiment three, a kind of device 700 of voice quality loss estimation, and the apparatus structure schematic diagram comprises as shown in Figure 7: radio link parameter extraction unit 710 and evaluation unit 720;
Radio link parameter extraction unit 710 is used for obtaining the radio link parameter of wireless signal;
Evaluation unit 720 is used for the parametric coefficients estimation voice quality loss of radio link parameter that obtains according to described radio link parameter extraction unit 710 and the estimation voice quality loss of presetting.
The embodiment of the invention four, a kind of device 800 of voice quality loss estimation, the apparatus structure schematic diagram comprises as shown in Figure 8: radio link parameter extraction unit 810, evaluation unit 820, described evaluation unit 820 comprises: parametric form converter unit 821 and loss evaluation unit 822.
Radio link parameter extraction unit 810 is used for obtaining the radio link parameter of wireless signal;
Parametric form converter unit 821 is used for described radio link parameter is transformed into the form relevant with the voice quality loss;
Loss evaluation unit 822 is according to the parametric coefficients estimation voice quality loss of described radio link parameter through form after the conversion and the estimation voice quality loss of presetting.
Described loss evaluation unit can be linear estimation device or non-linear estimation device or neural net.
The embodiment of the invention three and embodiment four described devices can move embodiment one described method.
Embodiment five, a kind of device 900 of estimating the parametric coefficients of voice quality loss, and the apparatus structure schematic diagram comprises as shown in Figure 9: radio link parameter acquiring unit 910, voice quality loss acquiring unit 920 and coefficient calculation unit 930;
Radio link parameter acquiring unit 910 is used to obtain the radio link parameter for the treatment of examining system;
Voice quality loss acquiring unit 920 is used for calculating according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal the voice quality loss value of radio link parameter correspondence;
Embodiment six, a kind of device of estimating the parametric coefficients of voice quality loss, the apparatus structure schematic diagram as shown in figure 10, comprise: radio link parameter acquiring unit 1010, voice quality loss acquiring unit 1020 and coefficient calculation unit 1030, described coefficient calculation unit comprises: parametric form converter unit 1031 and The Fitting Calculation unit 1032;
Radio link parameter acquiring unit 1010 is used to obtain the radio link parameter for the treatment of examining system;
Voice quality loss acquiring unit 1020 is used for calculating according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal the voice quality loss value of radio link parameter correspondence;
Parametric form converter unit 1031, the form relevant that is used for described radio link parameter is transformed into the voice quality loss;
The Fitting Calculation unit 1032 carries out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and described radio link parameter through the form after the conversion.
Described The Fitting Calculation unit is linear estimation device or non-linear estimation device or neural net.
The embodiment of the invention five and embodiment six described devices can move embodiment two described methods.
More than a kind of voice quality loss estimation method provided by the present invention and relevant apparatus are described in detail, wherein:
In the embodiment of the invention, obtain the radio link parameter in the wireless signal; And according to the parametric coefficients estimation voice quality loss of the described radio link parameter and the estimation voice quality loss of presetting.Relative variation by radio link parameter prediction wireless network transmissions voice, with respect to predicting the voice quality itself that receives in the prior art, can obtain better correlation, it is more accurate to predict, helps the voice transfer situation of wireless network is made correct evaluation.
For one of ordinary skill in the art, according to the thought of the embodiment of the invention, the part that all can change in specific embodiments and applications, in sum, this description should not be construed as limitation of the present invention.
Claims (14)
1. the evaluation method of a voice quality loss is characterized in that, comprising:
Obtain wireless signal;
Obtain the radio link parameter in the wireless signal, described radio link parameter comprises at least: delete frame per second;
Parametric coefficients estimation voice quality loss according to described radio link parameter and the estimation voice quality loss of presetting.
2. the method for claim 1 is characterized in that, described parametric coefficients estimation voice quality loss according to described radio link parameter and the estimation voice quality loss of presetting comprises:
Described radio link parameter is transformed into the form relevant with the voice quality loss;
According to of the parametric coefficients estimation voice quality loss of described radio link parameter through form after the conversion and the estimation voice quality loss of presetting.
3. method as claimed in claim 2 is characterized in that, the linear estimation of the process employing device of described estimation voice quality loss is carried out estimation or adopted non-linear estimation device execution estimation or utilize neural net to carry out estimation.
4. as claim 2 or 3 described methods, it is characterized in that, describedly described radio link parameter is transformed into the form relevant with the voice quality loss comprises:
By any one computing or the operation carried out in logarithm, index, power, evolution, quadrature, exponentiation or the weighted average windowing radio link parameter is carried out formal argument.
5. as any described method of claim 1 to 3, it is characterized in that described radio link parameter also comprises: the error rate, the longest wherein of deleting frame length, received signal level, soft information, speech energy parameter or handover statistic continuously.
6. as any described method of claim 1 to 3, it is characterized in that the parametric coefficients of described estimation voice quality loss is to obtain by following process The Fitting Calculation, comprising:
Obtain the radio link parameter for the treatment of examining system;
Calculate the voice quality loss value of radio link parameter correspondence according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal;
Carry out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter.
7. method as claimed in claim 6 is characterized in that, describedly carries out The Fitting Calculation according to described voice quality loss value and radio link parameter and obtains being used to estimating that the process of coefficient of each parameter of voice quality loss comprises:
The form relevant that described radio link parameter is transformed into the voice quality loss;
Carry out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter through the form after the conversion.
8. method as claimed in claim 7 is characterized in that, described The Fitting Calculation is that linear regression fit calculates or nonlinear prediction The Fitting Calculation or neural net calculating.
9. the device of a voice quality loss estimation is characterized in that, comprising:
The radio link parameter extraction unit is used for obtaining the radio link parameter of wireless signal, and described radio link parameter comprises at least: delete frame per second;
Evaluation unit is according to the parametric coefficients estimation voice quality loss of described radio link parameter and the estimation voice quality loss of presetting.
10. device as claimed in claim 9 is characterized in that, described evaluation unit comprises:
The parametric form converter unit is used for described radio link parameter is transformed into the form relevant with the voice quality loss;
The loss evaluation unit is according to the parametric coefficients estimation voice quality loss of described radio link parameter through form after the conversion and the estimation voice quality loss of presetting.
11. device as claimed in claim 10 is characterized in that, described loss evaluation unit is linear estimation device or non-linear estimation device or neural net.
12. a device of estimating the parametric coefficients of voice quality loss is characterized in that, comprising:
The radio link parameter acquiring unit is used to obtain the radio link parameter for the treatment of examining system, and described radio link parameter comprises at least: delete frame per second;
Voice quality loss acquiring unit is used for calculating according to the voice quality of the voice quality of given transmitting terminal and corresponding receiving terminal the voice quality loss value of radio link parameter correspondence;
Coefficient calculation unit is used for carrying out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and radio link parameter.
13. device as claimed in claim 12 is characterized in that, described coefficient calculation unit comprises:
The parametric form converter unit, the form relevant that is used for described radio link parameter is transformed into the voice quality loss;
The The Fitting Calculation unit carries out the coefficient that The Fitting Calculation obtains being used to estimating each parameter of voice quality loss according to described voice quality loss value and described radio link parameter through the form after the conversion.
14., it is characterized in that described The Fitting Calculation unit is linear estimation device or non-linear estimation device or neural net as claim 12 or 13 described devices.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007101724087A CN101459934B (en) | 2007-12-14 | 2007-12-14 | Voice quality loss estimation method and related apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2007101724087A CN101459934B (en) | 2007-12-14 | 2007-12-14 | Voice quality loss estimation method and related apparatus |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101459934A CN101459934A (en) | 2009-06-17 |
| CN101459934B true CN101459934B (en) | 2010-12-08 |
Family
ID=40770497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2007101724087A Expired - Fee Related CN101459934B (en) | 2007-12-14 | 2007-12-14 | Voice quality loss estimation method and related apparatus |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101459934B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102448083B (en) * | 2010-10-12 | 2016-01-20 | 中兴通讯股份有限公司 | A kind of radio communication voice quality assessment device, system and method |
| CN103024768B (en) * | 2011-09-23 | 2017-05-17 | 中兴通讯股份有限公司 | End-to-end videophone service quality evaluation method and system |
| CN104581758A (en) * | 2013-10-25 | 2015-04-29 | ***通信集团广东有限公司 | Voice quality estimation method and device as well as electronic equipment |
| CN110636176B (en) * | 2019-10-09 | 2022-05-17 | 科大讯飞股份有限公司 | Call fault detection method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1264526A (en) * | 1997-05-22 | 2000-08-23 | 艾利森电话股份有限公司 | Speech quality measurement in mobile telecommunication networks based on radio link parameters |
| CN1518300A (en) * | 2002-12-30 | 2004-08-04 | ���ǵ�����ʽ���� | Calling route selecting method based on predicted MOS |
| CN1523856A (en) * | 2002-12-25 | 2004-08-25 | �ձ����ŵ绰��ʽ���� | Estimation method and apparatus of overall conversational speech quality, program and recording medium for realizing the method |
-
2007
- 2007-12-14 CN CN2007101724087A patent/CN101459934B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1264526A (en) * | 1997-05-22 | 2000-08-23 | 艾利森电话股份有限公司 | Speech quality measurement in mobile telecommunication networks based on radio link parameters |
| CN1523856A (en) * | 2002-12-25 | 2004-08-25 | �ձ����ŵ绰��ʽ���� | Estimation method and apparatus of overall conversational speech quality, program and recording medium for realizing the method |
| CN1518300A (en) * | 2002-12-30 | 2004-08-04 | ���ǵ�����ʽ���� | Calling route selecting method based on predicted MOS |
| GB2398703A (en) * | 2002-12-30 | 2004-08-25 | Samsung Electronics Co Ltd | Call routing in voip based on MOS prediction value |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101459934A (en) | 2009-06-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108389592B (en) | Voice quality evaluation method and device | |
| Rix | Perceptual speech quality assessment-a review | |
| Uthansakul et al. | Estimating and synthesizing QoE based on QoS measurement for improving multimedia services on cellular networks using ANN method | |
| ATE363189T1 (en) | ESTIMATION OF TRANSMISSION QUALITY IN A RADIO NETWORK | |
| CN1321390C (en) | Establishment of statistics concerned model of acounstic quality normalization | |
| CN101459934B (en) | Voice quality loss estimation method and related apparatus | |
| EP1085534A3 (en) | Intelligent analysis system and method for fluid-filled electrical equipment | |
| CN102148033B (en) | Method for testing intelligibility of speech transmission index | |
| CN104581758A (en) | Voice quality estimation method and device as well as electronic equipment | |
| CN105679335B (en) | Speech quality assessment method and system based on no line analysis | |
| CN103607309A (en) | Mapping method for service KQI and QOE | |
| CN1523856A (en) | Estimation method and apparatus of overall conversational speech quality, program and recording medium for realizing the method | |
| CN101330703B (en) | Method, system and apparatus for detecting base station performance | |
| CN103561412B (en) | Channel associated shadow fading construction method based on stationary random process | |
| CN107666679A (en) | A kind of method and device of communication network alignment location algorithm parameter | |
| Van Wijngaarden et al. | THE SPEECH TRANSMISSION INDEX AFTER FOUR DECADES OF DEVELOPMENT. | |
| KR20080071611A (en) | Method, apparatus, analyser and computer program of searching for propagation paths | |
| DE50105867D1 (en) | Method for calibrating a test device | |
| EP1447940A2 (en) | Method for measuring a user perception score | |
| Wang et al. | A Novel $ A~ Priori $ Simulation Algorithm for Absorbing Receivers in Diffusion-Based Molecular Communication Systems | |
| Werner et al. | Quality control for AMR speech channels in GSM networks | |
| Bernardin et al. | Estimating the range to the cell edge from signal strength measurements | |
| JP4469278B2 (en) | How to evaluate whether time delay is better than test range | |
| CN103943114A (en) | Voice service conversation quality assessment method and apparatus | |
| CN110210711B (en) | Electric vehicle electric energy metering accuracy assessment method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20101208 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |