1 / 20

Standardisation supporting cultural diversity: From 5 to 28

This project aims to expand the language coverage of the ETSI spoken command vocabulary standard from 5 languages to 28. The standardization of spoken commands for ICT devices and services benefits users with limited access to terminals and displays, promotes e-inclusion and a multicultural Europe, and facilitates the greater uptake and use of ICTs. The methodology involves the generation of potential command words, confidence rating of command words, phonetic discrimination, and acoustic analysis. The timeline for the project spans from March 2006 to March 2008.

davidrogers
Télécharger la présentation

Standardisation supporting cultural diversity: From 5 to 28

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Standardisation supporting cultural diversity: From 5 to 28 Expanding the language coverage of the ETSI spoken command vocabulary standard. STF QD Mike Tate, Helge Hüttenrauch, Lutz Groh, Françoise Petersen Rosemary Orr,

  2. Advantages of Speech • Fundamental Human Paradigm, natural • Useful for people with limited access to terminals and displays • No age barrier • Safe • ASR technology commercially available • Becoming cheaper and more reliable

  3. Need for Standardisation • No previous standard - industry or other bodies • Increased prevalence of speech systems - therefore transfer of learning important • With enlarged EU comes increased national identity - therefore need to expand current standard • Facilitates e-inclusion and a multicultural Europe • Local language ASR systems increases local use - prevalence of English is a barrier • Greater uptake and use of ICTs - good for consumer QoL, service providers and equipment makers

  4. ETSI STF QD approach • Definition of languages covered • Definition of commands covered • Empirical data collection • Consultation and publication

  5. Generic spoken command vocabularyfor ICT devices and services • The ETSI Standard (ES) 202 076 specifies user tested spoken commands for five languages • French • German • Italian • Spanish • English ETSI STF QD

  6. Generic spoken command vocabularyfor ICT devices and services • Expansion of Standard to cover 28 Languages • Expansion of EU • Strong presence of Russia • EFTA Countries - Switzerland, Norway, Liechtenstein, Iceland ETSI STF QD

  7. Expansion of Standard to cover 28 Languages • Bulgarian, Czech, Croatian, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hungarian, Icelandic, Italian, Latvian, Lithuanian, Maltese, Norwegian, Polish, Portuguese, Romanian, Romansch, Russian, Slovak, Slovenian, Spanish, Swedish and Turkish.

  8. Current Voice Commands • Definition of commands covered: • Context-independent common commands (e.g. Options, Main menu, Standby) • Context-dependent common commands (e.g. Help, Repeat) • Core commands (e.g. Yes, No) • Digits • Name and digit dialling (e.g. Dial, Call) • Basic call handling and supplementary services (e.g. Answer, Busy)

  9. Current Voice Commands • Definition of commands covered: • Media control (e.g. Play, Pause) • Browseable list for navigation (e.g. Next, Previous) • Editing commands (e.g. Edit, Delete) • Device setting (e.g. Choose network, Sound off) • Word spotting mode (e.g. Wake up)

  10. Example Table

  11. Methodology 1. Spotaneous generation of potential command words • Storyboard method • Textual descriptions method 2. Confidence rating of command words • User testing of alternative commands 3. Phonetic discrimination • Recognizer field test • Pronunciation dictionary test

  12. Spontaneous generation of potential command words • Storyboard method • Illustrations by an artist - visual representation of functionality in combination with… • Background explanation • Participant responds with appropriate verbal description.

  13. Storyboard Method What does this represent?

  14. Spontaneous generation of potential command words • Textual descriptions method • Carefully worded description of functions • Participant responds with appropriate one-word description. • Descriptions sometimes verbose.

  15. Confidence rating of command words • Histogram frequency of words from initial data capture • Confidence rating from different participants on each alternative command name • Measured on a 5-point scale or… • Choose the command word for which they have most confidence.

  16. Accoustic Discrimination • Optimal discrimination between commands • Recogniser field test with ASR systems…but sensitive to • Different types of recognition system • Test database (recorded, live, styles) • 28 different recognisers required • Confusion matrix • Pronunciation dictionary test • Accoustic analysis of command words • Using lookup table of phone sequences

  17. Phonetic discriminability • ‘STOP’ - 4 phones • START - 5 phones But • ‘STOP’ - 2 Different phones • START - 3 Different phones In contrast • Begin- 5 phones • End - 3 phones And no common phones

  18. Timetable • Start of the work: March 2006 • Initial Draft ETSI Standard: September 2006 • Interim report to EC/EFTA: February 2007 • Final Draft ETSI Standard • Publication of ETSI Standard: March 2008

  19. Thank You! Mike Tate Chimera University of Essex mtate@essex.ac.uk

More Related