1、 ETSI TS 103 296 V1.1.1 (2016-08) Speech and Multimedia Transmission Quality (STQ); Requirements for Emotion Detectors used for Telecommunication Measurement Applications; Detectors for written text and spoken speech floppy3TECHNICAL SPECIFICATION ETSI ETSI TS 103 296 V1.1.1 (2016-08)2 Reference DTS
2、/STQ-236 Keywords internet, quality ETSI 650 Route des Lucioles F-06921 Sophia Antipolis Cedex - FRANCE Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16 Siret N 348 623 562 00017 - NAF 742 C Association but non lucratif enregistre la Sous-Prfecture de Grasse (06) N 7803/88 Important notice The present
3、 document can be downloaded from: http:/www.etsi.org/standards-search The present document may be made available in electronic versions and/or in print. The content of any electronic and/or print versions of the present document shall not be modified without the prior written authorization of ETSI.
4、In case of any existing or perceived difference in contents between such versions and/or in print, the only prevailing document is the print of the Portable Document Format (PDF) version kept on a specific network drive within ETSI Secretariat. Users of the present document should be aware that the
5、document may be subject to revision or change of status. Information on the current status of this and other ETSI documents is available at https:/portal.etsi.org/TB/ETSIDeliverableStatus.aspx If you find errors in the present document, please send your comment to one of the following services: http
6、s:/portal.etsi.org/People/CommiteeSupportStaff.aspx Copyright Notification No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm except as authorized by written permission of ETSI. The content of the PDF version shall not be
7、 modified without the written authorization of ETSI. The copyright and the foregoing restriction extend to reproduction in all media. European Telecommunications Standards Institute 2016. All rights reserved. DECTTM, PLUGTESTSTM, UMTSTMand the ETSI logo are Trade Marks of ETSI registered for the ben
8、efit of its Members. 3GPPTM and LTE are Trade Marks of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners. GSM and the GSM logo are Trade Marks registered and owned by the GSM Association. ETSI ETSI TS 103 296 V1.1.1 (2016-08)3 Contents Intellectual Property Right
9、s 6g3Foreword . 6g3Modal verbs terminology 6g3Introduction 6g31 Scope 7g32 References 7g32.1 Normative references . 7g32.2 Informative references 7g33 Definitions, symbols and abbreviations . 15g33.1 Definitions 15g33.2 Symbols 16g33.3 Abbreviations . 16g34 Emotion detectors for written text 18g34.1
10、 Introduction 18g34.2 Overview of emotion detectors 18g34.2.1 Introduction. 18g34.2.2 Current approaches . 18g34.2.3 Emotion detector description 21g34.3 Input . 21g34.4 Linguistic resources 21g34.4.1 Introduction. 21g34.4.2 Overview of resources 22g34.4.2.1 Description 22g34.4.2.2 WordNet 22g34.4.2
11、.3 Suggested Upper Merged Ontology (SUMO) . 22g34.4.2.4 Cyc database 22g34.4.2.5 Open Mind Common Sense (OMCS) . 23g34.4.2.6 ConceptNet 23g34.4.2.7 Other databases . 23g34.4.3 Annotation 23g34.5 Emotion models 24g34.5.1 Introduction. 24g34.5.2 Categorical emotion classification 24g34.5.3 Dimensional
12、 emotional classification . 25g34.6 Algorithms 28g34.6.1 Introduction. 28g34.6.2 Keyword based approach 29g34.6.2.1 Description 29g34.6.2.2 Advantages 29g34.6.2.3 Disadvantages . 29g34.6.2.4 Implementation . 29g34.6.2.5 Related works and results 30g34.6.3 Learning based approaches . 31g34.6.3.1 Desc
13、riptions 31g34.6.3.2 Support Vector Machines 31g34.6.3.2.1 Description 31g34.6.3.2.2 Advantages 32g34.6.3.2.3 Disadvantages 32g34.6.3.2.4 Related works and results 32g34.6.3.3 Nave Bayes Classifier 33g34.6.3.3.1 Description 33g34.6.3.3.2 Advantages 33g34.6.3.3.3 Disadvantages 33g34.6.3.3.4 Related w
14、orks and results 33g3ETSI ETSI TS 103 296 V1.1.1 (2016-08)4 4.6.3.4 Hidden Markov Model 33g34.6.3.4.1 Description 33g34.6.3.4.2 Advantages 34g34.6.3.4.3 Disadvantages 34g34.6.3.4.4 Related works and results 34g34.6.4 Hybrid approaches 34g34.6.4.1 Description 34g34.6.4.2 Related works and results 35g
15、34.7 Output . 35g34.7.1 Output description 35g34.7.2 Practical Examples 36g34.8 Final remarks on textual emotion detectors 39g35 Classification of emotions in scientific publications on speech recognition 40g35.1 Preface 40g35.2 Introduction 40g35.3 Basic information about speech emotions 42g35.4 La
16、nguage 43g35.5 Existing Corpora 44g35.5.1 Preface 44g35.5.2 Introduction. 44g35.5.3 English speech emotion databases 45g35.5.3.1 Preface . 45g35.5.3.2 Database 1 . 45g35.5.3.3 Database 2 . 45g35.5.3.4 Database 3 . 45g35.5.3.5 Database 4 . 45g35.5.3.6 Database 5 . 45g35.5.3.7 Database 6 . 45g35.5.3.8
17、 Database 7 . 46g35.5.3.9 Database 8 . 46g35.5.3.10 Database 9 . 46g35.5.3.11 Database 10 . 46g35.5.4 German speech emotion databases . 46g35.5.4.1 Preface . 46g35.5.4.2 Database 11 . 46g35.5.4.3 Database 12 . 46g35.5.4.4 Database 13 . 47g35.5.4.5 Database 14 . 47g35.5.4.6 Database 15 . 47g35.5.4.7
18、Database 16 . 47g35.5.4.8 Database 17 . 47g35.5.4.9 Database 18 . 47g35.5.5 Japanese speech emotion databases 47g35.5.5.1 Preface . 47g35.5.5.2 Database 19 . 47g35.5.5.3 Database 20 . 48g35.5.5.4 Database 21 . 48g35.5.6 Dutch emotion speech databases 48g35.5.6.1 Preface . 48g35.5.6.2 Database 22 . 4
19、8g35.5.6.3 Database 23 . 48g35.5.7 Spanish emotion speech databases 48g35.5.7.1 Preface . 48g35.5.7.2 Database 24 . 48g35.5.7.3 Database 25 . 49g35.5.8 Danish emotion speech database 49g35.5.8.1 Preface . 49g35.5.8.2 Database 26 . 49g35.5.9 Hebrew emotion speech database . 49g35.5.9.1 Preface . 49g3
20、5.5.9.2 Database 27 . 49g3ETSI ETSI TS 103 296 V1.1.1 (2016-08)5 5.5.10 Swedish emotion speech database 49g35.5.10.1 Preface . 49g35.5.10.2 Database 28 . 49g35.5.11 Chinese emotion speech database . 50g35.5.11.1 Preface . 50g35.5.11.2 Database 29 . 50g35.5.12 Russian emotion speech database . 50g35.
21、5.12.1 Preface . 50g35.5.12.2 Database 30 . 50g35.5.13 Multilingual emotion speech database 50g35.5.13.1 Preface . 50g35.5.13.2 Database 31 . 50g35.5.13.3 Database 32 . 50g35.5.14 Four other databases . 50g35.5.14.1 Preface . 50g35.5.14.2 Database 33 . 51g35.5.14.3 Database 34 . 51g35.5.14.4 Databas
22、e 35 . 51g35.5.14.5 Database 36 . 51g35.5.15 Summary . 51g35.6 Spoken Emotional Speech Pre-processing . 52g35.6.1 Preface 52g35.6.2 Introduction. 53g35.6.3 Features . 53g35.6.4 Parameters. 53g35.6.5 Methods and materials 54g36 Requirements for Emotion Detectors used for Telecommunication Measurement
23、 Applications and Systems. 54g36.1 General considerations . 54g36.1.1 Introduction. 54g36.1.2 Computational Power Requirements 57g36.1.3 Requirements for Operational Modes . 58g36.2 Requirements for Emotion Detectors for Written Text 58g36.3 Requirements for Emotion detectors for speech . 59g36.4 A
24、combined method and its requirements 61g36.4.1 General description . 61g36.4.2 Requirements of the combined method 61g37 Accuracy of Emotion Detectors for Subjective Testing in Telecommunications 62g37.1 Introduction 62g37.2 Reference set of samples 62g37.3 Assessment of the accuracy 62g37.4 Remaini
25、ng Percentage of Samples . 65g37.5 Examples of Single Emotion Detectors 68g37.6 Examples of combined emotion detectors 69g37.6.1 Optimum Recording process with no errors . 69g37.6.2 Poor Recording process with errors 73g3Annex A (informative): Overview of Available Speech Corpora . 75g3Annex B (info
26、rmative): Subjective Assessment of Emotional Content . 76g3Annex C (informative): Bibliography . 77g3History 85g3ETSI ETSI TS 103 296 V1.1.1 (2016-08)6 Intellectual Property Rights IPRs essential or potentially essential to the present document may have been declared to ETSI. The information pertain
27、ing to these essential IPRs, if any, is publicly available for ETSI members and non-members, and can be found in ETSI SR 000 314: “Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in respect of ETSI standards“, which is available from the ETSI Secretari
28、at. Latest updates are available on the ETSI Web server (https:/ipr.etsi.org/). Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on th
29、e ETSI Web server) which are, or may be, or may become, essential to the present document. Foreword This Technical Specification (TS) has been produced by ETSI Technical Committee Speech and multimedia Transmission Quality (STQ). Modal verbs terminology In the present document “shall“, “shall not“,
30、“should“, “should not“, “may“, “need not“, “will“, “will not“, “can“ and “cannot“ are to be interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression of provisions). “must“ and “must not“ are NOT allowed in ETSI deliverables except when used in direct citatio
31、n. Introduction It is important to mention that there is a difference between the concepts of sentiment and emotion. In the first case, it is expressed after deep thought and uses a well-organized lexicon. In the second case, it highly depends on specific situations and is expressed by physiological
32、 responses. Emotions are considered as a strong feeling while sentiment is mental attitude caused by feeling i.146. Hereinafter, these two terms will be represented as emotions and detectors dealing with these terms - emotion detectors. ETSI ETSI TS 103 296 V1.1.1 (2016-08)7 1 Scope The present docu
33、ment specifies Classification of and Requirements for Emotion Detectors for telecommunications and the assessment of their performance and uncertainties. 2 References 2.1 Normative references References are either specific (identified by date of publication and/or edition number or version number) o
34、r non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies. Referenced documents which are not found to be publicly available in the expected location might be found at http:/d
35、ocbox.etsi.org/Reference. NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity. The following referenced documents are necessary for the application of the present document. 1 ISO/IEC Guide 98-3:2008: “Uncertainty of
36、 measurement - Part 3: Guide to the expression of uncertainty in measurement (GUM:1995). 2 Recommendation ITU-T P.1401 (07/2012): “Methods, metrics and procedures for statistical evaluation, qualification and comparison of objective quality prediction models“. 3 Spiegel, M. (1998): “Theory and probl
37、ems of statistics“, McGraw Hill. 4 Recommendation ITU-T P.800 (08/1996): “Methods for subjective determination of transmission quality“. 2.2 Informative references References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific
38、 references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies. NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity. The fo
39、llowing referenced documents are not necessary for the application of the present document but they assist the user with regard to a particular subject area. i.1 P. Ekman: “Universals and cultural differences in facial expressions of emotion“ Nebraska symposium on motivation, pp. 207-282, 1972. i.2
40、P. Ekman, W. V Friesen, M. OSullivan, A. Chan, I. Diacoyanni-Tarlatzis, K. Heider, R. Krause, W. A. LeCompte, T. Pitcairn, P. E. Ricci-Bitti, K. Scherer and M. Tomita: “Universals and cultural differences in the judgments of facial expressions of emotion“, Journal of personality and social psycholog
41、y, vol. 53, no. 4. pp. 712-7, 1987. i.3 R. W. Picard: “Affective Computing“, Pattern Recognit., vol. 20, no. 321, p. 304, 1995. i.4 H. Atassi, M. T. Riviello, Z. Smkal, A. Hussain and A. Esposito: “Emotional Vocal Expressions Recognition Using the COST 2102 Italian Database of Emotional Speech“ in D
42、evelopment of Multimodal Interfaces Active Listening and Synchrony, 2010, pp. 255-267. ETSI ETSI TS 103 296 V1.1.1 (2016-08)8 i.5 H. Binali, C. Wu and V. Potdar: “Computational Approaches for Emotion Detection in Text“, 4thIEEE Int. Conf. Digit. Ecosyst. Technol. - Conf. Proc. IEEE-DEST 2010, DEST 2
43、010, vol. 37, no. 5, pp. 498-527, 2010. i.6 S. Gupta, A. Mehra and Vinay: “Speech emotion recognition using SVM with thresholding fusion“, 2ndInternational Conference on Signal Processing and Integrated Networks (SPIN), 2015, pp. 570-574. i.7 Y. Sun, C. Quan, X. Kang, Z. Zhang and F. Ren: “Customer
44、emotion detection by emotion expression analysis on adverbs“, Inf. Technol. Manag., vol. 16, no. 4, pp. 303-311, 2015. i.8 Y. Baimbetov, I. Khalil, M. Steinbauer and G. Anderst-Kotsis: “Using Big Data for Emotionally Intelligent Mobile Services through Multi-Modal Emotion Recognition“, Lecture Notes
45、 in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9102, A. Geissbhler, J. Demongeot, M. Mokhtari, B. Abdulrazak and H. Aloulou, Eds. Cham: Springer International Publishing, 2015, pp. 127-138. i.9 U. Krcadinac, P. Pasquier,
46、J. Jovanovic and V. Devedzic: “Synesketch: An Open Source Library for Sentence-Based Emotion Recognition“, IEEE Trans. Affect. Comput., vol. 4, no. 3, pp. 312-325, 2013. i.10 A. Neviarouskaya, H. Prendinger and M. Ishizuka: “Affect Analysis Model: novel rule-based approach to affect sensing from tex
47、t“ Nat. Lang. Eng., vol. 17, no. 01, pp. 95-135, Sep. 2011. i.11 S. Aman and S. Szpakowicz: “Identifying Expressions of Emotion in Text“ in Text, Speech and Dialogue, vol. 4629, Berlin, Heidelberg: Springer Berlin Heidelberg, 2007, pp. 196-205. i.12 G. Valkanas and D. Gunopulos: “A UI Prototype for
48、Emotion-Based Event Detection in the Live Web“ Human-Computer Interact. Knowl. Discov. Complex, Unstructured, Big Data, vol. 7947 LNCS, pp. 89-100, 2013. i.13 S. Shaheen, W. El-Hajj, H. Hajj and S. Elbassuoni: “Emotion Recognition from Text Based on Automatically Generated Rules“ in 2014 IEEE Intern
49、ational Conference on Data Mining Workshop, 2014, pp. 383-392. i.14 C. Ma, H. Prendinger and M. Ishizuka: “Emotion Estimation and Reasoning Based on Affective Textual Interaction“, Lecture Notes Comput. Sci. (including Subser. Lecture Notes Artif. Intell. and Lecture Notes Bioinformatics), vol. 3784 LNCS, pp. 622-628, 2005. i.15 D. T. Ho and T. H. Cao: “A High-Order Hidden Markov Model for Emotion Detection from Textual Data“, Lecture Notes in Computer Science (Knowledge Management and Acquisition for Intelligent Systems), vol. 74
