ITU-T P 862-2001 Perceptual evaluation of speech quality (PESQ) An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs--SEN.pdf

《ITU-T P 862-2001 Perceptual evaluation of speech quality (PESQ) An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs--SEN.pdf》由会员分享，可在线阅读，更多相关《ITU-T P 862-2001 Perceptual evaluation of speech quality (PESQ) An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs--SEN.pdf（30页珍藏版）》请在麦多课文档分享上搜索。

1、 INTERNATIONAL TELECOMMUNICATION UNION ITU-T P.862TELECOMMUNICATION STANDARDIZATION SECTOR OF ITU (02/2001) SERIES P: TELEPHONE TRANSMISSION QUALITY, TELEPHONE INSTALLATIONS, LOCAL LINE NETWORKS Methods for objective and subjective assessment of quality Perceptual evaluation of speech quality (PESQ)

2、: An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs ITU-T Recommendation P.862 (Formerly CCITT Recommendation) ITU-T P-SERIES RECOMMENDATIONS TELEPHONE TRANSMISSION QUALITY, TELEPHONE INSTALLATIONS, LOCAL LINE NETWORKS Vocabulary and eff

3、ects of transmission parameters on customer opinion of transmission quality Series P.10 Subscribers lines and sets Series P.30 P.300 Transmission standards Series P.40 Objective measuring apparatus Series P.50 P.500 Objective electro-acoustical measurements Series P.60 Measurements related to speech

4、 loudness Series P.70 Methods for objective and subjective assessment of quality Series P.80 P.800 Audiovisual quality in multimedia services Series P.900 For further details, please refer to the list of ITU-T Recommendations. ITU-T Recommendation P.862 Perceptual evaluation of speech quality (PESQ)

5、: An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs Summary This Recommendation describes an objective method for predicting the subjective quality of 3.1 kHz (narrow-band) handset telephony and narrow-band speech codecs. This Recommenda

6、tion presents a high-level description of the method, advice on how to use it, and part of the results from a Study Group 12 benchmark carried out in the period 1999-2000. An ANSI-C reference implementation, described in Annex A, is provided in separate files and form an integral part of this Recomm

7、endation. A conformance testing procedure is also specified in Annex A to allow a user to validate that an alternative implementation of the model is correct. This ANSI-C reference implementation shall take precedence in case of conflicts between the high-level description as given in this Recommend

8、ation and the ANSI-C reference implementaion. This Recommendation includes an electornic attachment containing an ANSI-C reference implementation of PESQ and conformance testing data. Source ITU-T Recommendation P.862 was prepared by ITU-T Study Group 12 (2001-2004) and approved under the WTSA Resol

9、ution 1 procedure on 23 February 2001. ITU-T P.862 (02/2001) i FOREWORD The International Telecommunication Union (ITU) is the United Nations specialized agency in the field of telecommunications. The ITU Telecommunication Standardization Sector (ITU-T) is a permanent organ of ITU. ITU-T is responsi

10、ble for studying technical, operating and tariff questions and issuing Recommendations on them with a view to standardizing telecommunications on a worldwide basis. The World Telecommunication Standardization Assembly (WTSA), which meets every four years, establishes the topics for study by the ITU-

11、T study groups which, in turn, produce Recommendations on these topics. The approval of ITU-T Recommendations is covered by the procedure laid down in WTSA Resolution 1. In some areas of information technology which fall within ITU-Ts purview, the necessary standards are prepared on a collaborative

12、basis with ISO and IEC. NOTE In this Recommendation, the expression “Administration“ is used for conciseness to indicate both a telecommunication administration and a recognized operating agency. INTELLECTUAL PROPERTY RIGHTS ITU draws attention to the possibility that the practice or implementation

13、of this Recommendation may involve the use of a claimed Intellectual Property Right. ITU takes no position concerning the evidence, validity or applicability of claimed Intellectual Property Rights, whether asserted by ITU members or others outside of the Recommendation development process. As of th

14、e date of approval of this Recommendation, ITU had received notice of intellectual property, protected by patents, which may be required to implement this Recommendation. However, implementors are cautioned that this may not represent the latest information and are therefore strongly urged to consul

15、t the TSB patent database. ITU 2001 All rights reserved. No part of this publication may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm, without permission in writing from ITU. ii ITU-T P.862 (02/2001) CONTENTS Page 1 Introductio

16、n 1 2 Normative references 1 3 Abbreviations . 1 4 Scope 2 5 Conventions 4 6 Overview of PESQ . 4 7 Comparison between objective and subjective scores 6 7.1 Correlation coefficient 6 7.2 Residual errors 6 8 Preparation of processed speech material. 7 8.1 Source material. 7 8.1.1 Choice of source mat

17、erial 7 8.1.2 ITU-T Temporal structure and duration of source material 7 8.1.3 Filtering and level calibration . 8 8.2 Addition of background noise 8 8.3 Processing through system under test. 8 9 Selection of experimental parameters 8 10 Description of PESQ algorithm 9 10.1 Level and time alignment

18、pre-processing (Figure 3) . 13 10.1.1 Computation of the overall system gain 13 10.1.2 IRS filtering. 13 10.1.3 Time alignment . 13 10.2 Perceptual model (Figures 4a and 4b) 15 10.2.1 Precomputation of constant settings 15 10.2.2 IRS-receive filtering 15 10.2.3 Computation of the active speech time

19、interval 15 10.2.4 Short-term Fast Fourier Transform . 16 10.2.5 Calculation of the pitch power densities . 16 10.2.6 Partial compensation of the original pitch power density for transfer function equalization . 16 10.2.7 Partial compensation of the distorted pitch power density for time-varying gai

20、n variations between distorted and original signal . 16 10.2.8 Calculation of the loudness densities 16 10.2.9 Calculation of the disturbance density 17 10.2.10 Cell-wise multiplication with an asymmetry factor 17 ITU-T P.862 (02/2001) iii Page 10.2.11 Aggregation of the disturbance densities over f

21、requency and emphasis on soft parts of the original 17 10.2.12 Zeroing of the frame disturbance for frames during which the delay decreased significantly 18 10.2.13 Realignment of bad intervals 18 10.2.14 Aggregation of the disturbance within split second intervals 18 10.2.15 Aggregation of the dist

22、urbance over the duration of the speech signal (around 10 s), including a recency factor 18 10.2.16 Computation of the PESQ score. 18 Annex A Reference implementation of PESQ and conformance testing . 19 Electronic attachment: ANSI-C reference implementation of PESQ and conformance testing data. iv

23、ITU-T P.862 (02/2001) ITU-T Recommendation P.862 Perceptual evaluation of speech quality (PESQ): An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs11 Introduction The objective method described in this Recommendation is known as “Perceptu

24、al Evaluation of Speech Quality“ (PESQ). It is the result of several years of development and is applicable not only to speech codecs but also to end-to-end measurements. Real systems may include filtering and variable delay, as well as distortions due to channel errors and low bit-rate codecs. The

25、PSQM method as described in ITU-T P.861 (February 1998), was only recommended for use in assessing speech codecs, and was not able to take proper account of filtering, variable delay, and short localized distortions. PESQ addresses these effects with transfer function equalization, time alignment, a

26、nd a new algorithm for averaging distortions over time. The validation of PESQ included a number of experiments that specifically tested its performance across combinations of factors such as filtering, variable delay, coding distortions and channel errors. It is recommended that PESQ be used for sp

27、eech quality assessment of 3.1 kHz (narrow-band) handset telephony and narrow-band speech codecs. 2 Normative references The following ITU-T Recommendations and other reference contain provisions which, through reference in this text, constitute provisions of this Recommendation. At the time of publ

28、ication, the editions indicated were valid. All Recommendations and other references are subject to revision; users of this Recommendation are therefore encouraged to investigate the possibility of applying the most recent edition of the Recommendations and other references listed below. A list of t

29、he currently valid ITU-T Recommendations is regularly published. ITU-T P.800 (1996), Methods for subjective determination of transmission quality. ITU-T P.810 (1996), Modulated noise reference unit (MNRU). ITU-T P.830 (1996), Subjective performance assessment of telephone-band and wideband digital c

30、odecs. ITU-T P-series Supplement 23 (1998), ITU-T coded-speech database. 3 Abbreviations This Recommendation uses, the following abbreviations: ACR Absolute Category Rating CELP Code-Excited Linear Prediction DMOS Degradation Mean Opinion Score HATS Head And Torso Simulator _ 1This Recommendation in

31、cludes an electronic attachment containing an ANSI-C reference implementation of PESQ and conformance testing data. ITU-T P.862 (02/2001) 1 IRS Intermediate Reference System LQ Listening Quality MOS Mean Opinion Score PCM Pulse Code Modulation PESQ Perceptual Evaluation of Speech Quality PSQM Percep

32、tual Speech Quality Measure 4 Scope Based on the benchmark results presented within Study Group 12, an overview of the test factors, coding technologies and applications to which this Recommendation applies is given in Tables 1 to 3. Table 1 presents the relationships of test factors, coding technol

33、ogies and applications for which this Recommendation has been found to show acceptable accuracy. Table 2 presents a list of conditions for which the Recommendation is known to provide inaccurate predictions or is otherwise not intended to be used. Finally, Table 3 lists factors, technologies and app

34、lications for which PESQ has not currently been validated. Although correlations between objective and subjective scores in the benchmark were around 0.935 for both known and unknown data, the PESQ algorithm cannot be used to replace subjective testing. It should also be noted that the PESQ algorith

35、m does not provide a comprehensive evaluation of transmission quality. It only measures the effects of one-way speech distortion and noise on speech quality. The effects of loudness loss, delay, sidetone, echo, and other impairments related to two-way interaction (e.g. centre clipper) are not reflec

36、ted in the PESQ scores. Therefore, it is possible to have high PESQ scores, yet poor quality of the connection overall. Table 1/P.862 Factors for which PESQ had demonstrated acceptable accuracy Test factors Speech input levels to a codec Transmission channel errors Packet loss and packet loss concea

37、lment with CELP codecs Bit rates if a codec has more than one bit-rate mode Transcodings Environmental noise at the sending side (See Note.) Effect of varying delay in listening only tests Short-term time warping of audio signal Long-term time warping of audio signal Coding technologies Waveform cod

38、ecs, e.g. G.711; G.726; G.727 CELP and hybrid codecs 4 kbit/s, e.g. G.728, G.729, G.723.1 Other codecs: GSM-FR, GSM-HR, GSM-EFR, GSM-AMR, CDMA-EVRC, TDMA-ACELP, TDMA-VSELP, TETRA 2 ITU-T P.862 (02/2001) Applications Codec evaluation Codec selection Live network testing using digital or analogue conn

39、ection to the network Testing of emulated and prototype networks NOTE When environmental noise is present the quality can be measured by passing PESQ the clean original without noise, and the degraded signal with noise. Table 2/P.862 PESQ is known to provide inaccurate predictions when used in conju

40、nction with these variables, or is otherwise not intended to be used with these variables Test factors Listening levels (See Note.) Loudness loss Effect of delay in conversational tests Talker echo Sidetone Coding technologies Replacement of continuous sections of speech making up more than 25% of a

41、ctive speech by silence (extreme temporal clipping) Applications In-service non-intrusive measurement devices Two-way communications performance NOTE PESQ assumes a standard listening level of 79 dB SPL and compensates for non-optimum signal levels in the input files. The subjective effect of deviat

42、ion from optimum listening level is therefore not taken into account. Table 3/P.862 (For further study) Factors, technologies and applications for which PESQ has not currently been validated Test factors Packet loss and packet loss concealment with PCM type codecs (See Note 1.) Temporal clipping of

43、speech (See Note 1.) Amplitude clipping of speech (See Note 2.) Talker dependencies Multiple simultaneous talkers Bit-rate mismatching between an encoder and a decoder if a codec has more than one bit-rate mode Network information signals as input to a codec Artificial speech signals as input to a c

44、odec Music as input to a codec ITU-T P.862 (02/2001) 3 Table 3/P.862 (For further study) Factors, technologies and applications for which PESQ has not currently been validated Test factors Listener echo Effects/artifacts from operation of echo cancellers Effects/artifacts from noise reduction algori

45、thms Coding technologies CELP and hybrid codecs 4 kbit/s MPEG4 HVXC Applications Acoustic terminal/handset testing, e.g. using HATS NOTE 1 PESQ appears to be more sensitive than subjects to front-end temporal clipping, especially in the case of missing words which may not be perceived by subjects. C

46、onversely, PESQ may be less sensitive than subjects to regular, short time clipping (replacement of short sections of speech by silence). In both of these cases there may be reduced correlation between PESQ and subjective MOS. NOTE 2 There is some evidence to suggest that PESQ is able to account for

47、 amplitude clipping, but only four conditions are known to have been included (in two 50-condition experiments) in the validation database described in clause 7. 5 Conventions Subjective evaluation of telephone networks and speech codecs may be conducted using listening-only or conversational method

48、s of subjective testing. For practical reasons, listening-only tests are the only feasible method of subjective testing during the development of speech codecs, when a real-time implementation of the codec is not available. This Recommendation discusses an objective measurement technique for estimat

49、ing subjective quality obtained in listening-only tests, using listening equipment conforming to the IRS or modified IRS receive characteristics. Most information on the performance of PESQ is from ACR listening quality (LQ) subjective experiments. This Recommendation should therefore be considered to relate primarily to the ACR LQ opinion scale. 6 Overview of PESQ PESQ compares an original signal X(t) with a degraded signal Y(t) that is the result of passing X(t) through a communications system. The output of PESQ is a prediction o