ITU-T P 85 AMD 1-2013 A method for subjective performance assessment of the quality of speech voice output devices Amendment 1 New Appendix I C Evaluation of speech output for audi.pdf

《ITU-T P 85 AMD 1-2013 A method for subjective performance assessment of the quality of speech voice output devices Amendment 1 New Appendix I C Evaluation of speech output for audi.pdf》由会员分享，可在线阅读，更多相关《ITU-T P 85 AMD 1-2013 A method for subjective performance assessment of the quality of speech voice output devices Amendment 1 New Appendix I C Evaluation of speech output for audi.pdf（14页珍藏版）》请在麦多课文档分享上搜索。

1、 International Telecommunication Union ITU-T P.85TELECOMMUNICATION STANDARDIZATION SECTOR OF ITU Amendment 1(03/2013) SERIES P: TERMINALS AND SUBJECTIVE AND OBJECTIVE ASSESSMENT METHODS Methods for objective and subjective assessment of speech quality A method for subjective performance assessment o

2、f the quality of speech voice output devices Amendment 1: New Appendix I Evaluation of speech output for audiobook reading tasks Recommendation ITU-T P.85 (1994) Amendment 1 ITU-T P-SERIES RECOMMENDATIONS TERMINALS AND SUBJECTIVE AND OBJECTIVE ASSESSMENT METHODS Vocabulary and effects of transmissio

3、n parameters on customer opinion of transmission quality Series P.10 Voice terminal characteristics Series P.30 P.300 Reference systems Series P.40 Objective measuring apparatus Series P.50 P.500 Objective electro-acoustical measurements Series P.60 Measurements related to speech loudness Series P.7

4、0 Methods for objective and subjective assessment of speech quality Series P.80P.800Audiovisual quality in multimedia services Series P.900 Transmission performance and QoS aspects of IP end-points Series P.1000 Communications involving vehicles Series P.1100 Models and tools for quality assessment

5、of streamed media Series P.1200 Telemeeting assessment Series P.1300 Statistical analysis, evaluation and reporting guidelines of quality measurements Series P.1400 For further details, please refer to the list of ITU-T Recommendations. Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) i Recommendation ITU-T P

6、.85 A method for subjective performance assessment of the quality of speech voice output devices Amendment 1 New Appendix I Evaluation of speech output for audiobook reading tasks Summary Amendment 1 to Recommendation ITU-T P.85 introduces a test methodology that addresses the evaluation of speech o

7、utput for audiobook reading tasks. History Edition Recommendation Approval Study Group 1.0 ITU-T P.85 1994-06-21 12 1.1 ITU-T P.85 (1994) Amd. 1 2013-03-28 12 ii Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) FOREWORD The International Telecommunication Union (ITU) is the United Nations specialized agency i

8、n the field of telecommunications, information and communication technologies (ICTs). The ITU Telecommunication Standardization Sector (ITU-T) is a permanent organ of ITU. ITU-T is responsible for studying technical, operating and tariff questions and issuing Recommendations on them with a view to s

9、tandardizing telecommunications on a worldwide basis. The World Telecommunication Standardization Assembly (WTSA), which meets every four years, establishes the topics for study by the ITU-T study groups which, in turn, produce Recommendations on these topics. The approval of ITU-T Recommendations i

10、s 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 basis with ISO and IEC. NOTE In this Recommendation, the expression “Administration“ is used for conciseness to

11、 indicate both a telecommunication administration and a recognized operating agency. Compliance with this Recommendation is voluntary. However, the Recommendation may contain certain mandatory provisions (to ensure, e.g., interoperability or applicability) and compliance with the Recommendation is a

12、chieved when all of these mandatory provisions are met. The words “shall“ or some other obligatory language such as “must“ and the negative equivalents are used to express requirements. The use of such words does not suggest that compliance with the Recommendation is required of any party. INTELLECT

13、UAL PROPERTY RIGHTS ITU draws attention to the possibility that the practice or implementation 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, whethe

14、r asserted by ITU members or others outside of the Recommendation development process. As of the date of approval of this Recommendation, ITU had not received notice of intellectual property, protected by patents, which may be required to implement this Recommendation. However, implementers are caut

15、ioned that this may not represent the latest information and are therefore strongly urged to consult the TSB patent database at http:/www.itu.int/ITU-T/ipr/. ITU 2013 All rights reserved. No part of this publication may be reproduced, by any means whatsoever, without the prior written permission of

16、ITU. Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) iii Table of Contents Page Amendment 1 New Appendix I Evaluation of speech output for audiobook reading tasks 1 I.1 Speech material 2 I.2 Rating scales . 2 I.3 Test procedure 4 Bibliography. 5 Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) 1 Recommendation ITU-T P

17、.85 A method for subjective performance assessment of the quality of speech voice output devices Amendment 1 New Appendix I Evaluation of speech output for audiobook reading tasks (This appendix does not form an integral part of the Recommendation.) Whereas the method and the questionnaires given in

18、 the main body of this Recommendation are adequate for applications providing vocal answers related to telephone directory inquiries, weather forecast, mail order and similar tasks, they are less adequate for applications where longer text paragraphs and potentially literature are read through synth

19、etic speech output, as is the case in audiobook reading tasks. For such applications, the task of the voice output is not pure information provisioning, but rather an expressive, emotion-triggering vocal output, putting more emphasis on different aspects of speech output quality than in traditional

20、telecom services 9. To address these aspects, a test methodology is presented in this Appendix which addresses mainly two perceptual dimensions of such speech output, namely its listening pleasure and its prosody. These dimensions were extracted with the help of principal axis factor (PAF) analysis

21、from auditory test results involving typical examples of TTS systems and typical (prosaic and poetic) text passages, as summarized in Table I.1. Details on the derivation of the methodology are given in 8 and 10. Table I.1 Examples of text categories for audiobook reading tasks ID Category Author Bo

22、ok 1 Long sentences Sven Regener Der kleine Bruder12 Direct speech, incomplete sentences Douglas Adams The Hitchhikers Guide to the Galaxy 3 Higher level of lexis, complex sentence structure Charles Dickens The Adventures of Oliver Twist 4 Poetic, picturesque Antoine de Saint-Exupry Wind, Sand and S

23、tars 5 Direct speech, basic language Tommy Jaud Resturlaub16 Action, short sentences Thomas Harris Hannibal 7 Childrens book Astrid Lindgren Pippi Longstocking 8 Thriller Ken Follett Code to Zero In the following paragraphs, the differences from the standard procedure described in the main body of t

24、his Recommendation are outlined. All other characteristics of the test methodology (selection of test participants, listening environment, etc.) should remain as described previously. _ 1No English translation available. 2 Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) I.1 Speech material Passages from stan

25、dard audio books should be used as material for the listening test. The passages should cover a wide variety of writing styles and book categories, potentially including thrillers, funny books, action-packed passages, books for children, books with very long sentences, and passages containing almost

26、 only direct speech, in case that these styles are considered as representative for the later application. Exemplary text categories are given in Table I.1. The passages should have an approximate length of one minute when read aloud. I.2 Rating scales For assessing speech output in audiobook readin

27、g tasks, a questionnaire with eight items is proposed, see Table I.2. Four of these items are modified versions of items in the main body of this Recommendation, and four items are added which are specific to audiobook reading tasks. These additional items were selected based on the review of curren

28、t literature, and address prosodic elements like communicative, structuring, aesthetic and emotional aspects that can be seen as the most important factors for reading and interpreting books 11. Table I.2 Questionnaire items for audiobook reading tasks Standard items (modified) Additional items rela

29、ted to audiobooks Overall impression Speech pauses Voice pleasantness Intonation Listening effort Emotion Acceptance Word stress The scales address the following aspects of quality: Overall impression This scale evaluates the overall quality of the synthesized signal and was adopted with slight modi

30、fications. Voice pleasantness This scale measures the degree of voice pleasantness from unpleasant to pleasant and was adopted with slight modifications. Listening effort This scale describes the effort a listener is required to make when listening to this voice over a longer period of time. It was

31、adopted from this Recommendation with slight modifications. Acceptance The acceptance scale from the main body of this Recommendation was transformed into a continuous scale. Speech pauses This scale evaluates if punctuation marks (e.g., period, comma, question mark, exclamation mark, colon, etc.) h

32、ave been converted into appropriate speech pauses between words, sentences and paragraphs. Intonation This scale determines if the produced pitch curve fits with the sentence type, e.g., the pitch of interrogative sentences usually increases at the end of a sentence whereas the pitch of declarative

33、sentences decreases. Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) 3 Emotion Variation of emotion is achieved by variations of sound pressure, intonation, speech pauses and volume 11. To ensure an authentic reading experience, the voice should reflect the atmosphere of the scene and the moods of the charac

34、ters. Word stress Unnatural stress and accentuations often result in very annoying voices and thus also cause problems in text comprehension 11. This scale is used to determine if the stress was perceived as unnatural or confusing. The scales are presented in a continuous scale layout with overflow

35、areas on both extremities, in order to avoid saturation effects. Such effects might occur if a characteristic of one stimulus is already rated in an end category of a scale, and the subsequent stimulus is considered as having an even more extreme characteristic of that kind. The scales may be presen

36、ted on paper (by collecting responses with a pencil), or on a graphical user interface (by collecting responses with sliders on a computer screen). P.85-Amd.1(13)_FI.01EmotionDid you think the voice expressed an appropriate emotion for this text?Word stressWhat did you think of the way words in the

37、passage were stressed?Stressunnatural/confusingMelody fitted the sentence typeUnnatural/no expression of emotionsAuthentic expression of emotionsIntonationWhat did you think of the “melody” of the voice reading this passage? Speech pausesappropriate/pleasantSpeech pausesconfusing/unpleasantSuitable,

38、excellent realizationMelody did not fit the sentence typeListening effortHow would you describe the effort to listen to this voice over a longer period of time?Speech pausesHow did the pauses between words and sentences affect your listening to the passage?Very exhaustingNot suitable, bad realizatio

39、nAcceptanceDo you think that this voice could be used for synthesizing this audiobook?Very easyVery unpleasant Very pleasantOverall impressionHow do you rate the overall quality of the sound considering all aspects?Voice pleasantnessHow pleasant did you find the voice you just heard?Bad ExcellentStr

40、essnaturalFigure I.1 Graphical layout of rating scales for audiobook reading tasks 4 Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) I.3 Test procedure Test participants are instructed to first rate their overall impressions of the stimulus on the continuous rating scale ranging from bad to excellent. Subseq

41、uently, quality estimates for the other eight scales are solicited. To avoid any impact with regard to the order, the sequence of these subsequent eight scales (except MOS) should be randomized between participants. In order to familiarize themselves with the test procedure, participants first have

42、to pass a training phase with approximately two stimuli that are not included in the main test. Rec. ITU-T P.85 (1994)/Amd.1 (03/2013) 5 Bibliography 8 F. Hinterleitner, G. Neitzel, S. Mller, C. Norrenbrock, “An Evaluation Protocol for the Subjective Assessment of Text-to-Speech in Audiobook Reading

43、 Tasks“, in: Blizzard Challenge 2011 Workshop, Sept. 2, Torino, 2011. http:/citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.227.8997 9 F. Hinterleitner, S. Mller, C. Norrenbrock, and U. Heute, “Perceptual Quality Dimensions of Text-to-Speech Systems“, in: Proceedings of the 12th Annual Conference o

44、f the ISCA (Interspeech 2011), International Speech Communication Association (ISCA), 2011. http:/www.interspeech2011.org/conference/programme/sessionlist-static.html 10 F. Hinterleitner, C. Norrenbrock, and S. Mller, “Perceptual Quality Dimensions of Text-to-Speech Systems in Audiobook Reading Task

45、s“, in: Proceedings of the 24. Konferenz zur Elektronischen Sprachsignalverarbeitung (ESSV 2013), Bielefeld, 2013. 11 U. Rautenberg and T. Schnickmann, Das Hrbuch Stimme und Inszenierung. Harrassowitz Verlag, Wiesbaden, 2007, Chapter “Die Stimme im Hrbuch: Literaturverlust oder Sinnlichkeitsgewinn?“

46、, pp. 21-54. Printed in Switzerland Geneva, 2013 SERIES OF ITU-T RECOMMENDATIONS Series A Organization of the work of ITU-T Series D General tariff principles Series E Overall network operation, telephone service, service operation and human factors Series F Non-telephone telecommunication services

47、Series G Transmission systems and media, digital systems and networks Series H Audiovisual and multimedia systems Series I Integrated services digital network Series J Cable networks and transmission of television, sound programme and other multimedia signals Series K Protection against interference

48、 Series L Construction, installation and protection of cables and other elements of outside plant Series M Telecommunication management, including TMN and network maintenance Series N Maintenance: international sound programme and television transmission circuits Series O Specifications of measuring

49、 equipment Series P Terminals and subjective and objective assessment methods Series Q Switching and signalling Series R Telegraph transmission Series S Telegraph services terminal equipment Series T Terminals for telematic services Series U Telegraph switching Series V Data communication over the telephone network Series X Data networks, open system communications and security Series Y Global information infrastructure, Internet protocol aspects and next-generation networks Series Z Languages and gener