Dave Toney, Korin Richmond Toney, Korin Richmond Dave David - PowerPoint PPT Presentation

Dave Toney, Korin Richmond Toney, Korin Richmond Dave David Feinberg David Feinberg Edinburgh University Edinburgh University St Andrews University St Andrews University

travel/traffic info directions weather restaurant theatre/concerts accommodation

gender gender improved recognition improved recognition age age consumer preferences consumer preferences accent accent interface complexity interface complexity emotion emotion disposable income disposable income (location) (location) safety safety

S: Welcome to the Glasgow automated tourist information line. the Glasgow automated tourist information line. S: Welcome to How can I help you? How can I help you? U: Are there any museums nearby? U: Are there any museums nearby? S: The Gallery of Modern Art is closest to you. Admission is free. The Gallery of Modern Art is closest to you. Admission is free. S: U: Are there any plays on tonight? Are there any plays on tonight? U: S: : Yes, Macbeth is showing at 6pm in Botanic Park. S Yes, Macbeth is showing at 6pm in Botanic Park. You should probably get a taxi or use the underground. You should probably get a taxi or use the underground. U: OK, what s the weather like this evening? U: OK, what s the weather like this evening? S: : Initially fine, with rain expected later. S Initially fine, with rain expected later. U: U: Oh, right. Anything else on? Oh, right. Anything else on?

pitch, jitter shimmer harmonics - to - noise ratio

microphone mobile

ACOUSTI C NEURAL FEATURE NETWORK EXTRACTOR SPEECH DATABASE

.wav ACOUSTI C NEURAL FEATURE NETWORK EXTRACTOR gender, age

Acoustic Feature Extractor: Praat Praat Acoustic Feature Extractor: Neural Network: Netlab (Matlab) Netlab (Matlab) Neural Network: CTIMIT: 3303 recordings, 621 speakers 3303 recordings, 621 speakers CTIMIT: 429 male, 192 female 429 male, 192 female age 21 - - 55 55 age 21

Gender - - 94.4% accuracy 94.4% accuracy Gender Age - - no useful learning, possible reasons: no useful learning, possible reasons: Age little variation little variation background noise background noise not enough data not enough data

Gender - - 94.4% accuracy 94.4% accuracy Gender Age - - no useful learning, possible reasons: no useful learning, possible reasons: Age little variation little variation background noise background noise MACROPHONE (land-line) not enough data not enough data

Conversational interfaces Conversational interfaces Personalizing interaction very worthwhile Personalizing interaction very worthwhile Acoustic cues to user profile Acoustic cues to user profile Integrated profiling module Integrated profiling module

gender gender improved recognition improved recognition age age consumer preferences consumer preferences accent accent interface complexity interface complexity emotion emotion disposable income disposable income height height safety safety dave@cstr.ed.ac.uk