AVOZES

The Audio-Video Australian English Speech Data Corpus

Contents Module 3 - 'Calibration' sequences

This module comprises two sequences per speaker for the purpose of `speaker calibration', in terms of their visible speech articulation or visual expressiveness. For (purely visual) lipreading as well as AV automatic speech recognition, the amount of visible speech articulation determines how much (additional) information can possibly be gained from the video stream. Expressive visible speech articulation offers more information than a person who does not move the visible speech articulators much (for example, a person who mumbles). Extracting lip parameters, such as mouth width or mouth height, over time enables an analysis of the visual expressiveness of a speaker, for example by analysing the maximum values reached in each cycle of lip movements. Speakers with values in the margin of the overall distribution can be excluded from the analysis or treated differently, if desired.

The two calibration sequences "ba ba ba ..." (/bɑː bɑː bɑː .../) and "e o e o e o ..." (/iː ɔː iː ɔː iː ɔː .../) recorded in the AVOZES data corpus were each repeated continuously by each speaker for about 10 seconds. Despite the artificial nature of these prompts, the first sequence can give insight into the amount of vertical lip movement, i.e. opening and closing, while the second sequence emphasises horizontal lip movement, i.e. rounding and stretching.

Note: If your browser does not show the IPA symbols above correctly, please select a Unicode font.

Example Sequence
Note: Any example sequence is provided for informative purposes, so that you can judge whether AVOZES is the right data corpus for you. You may use it for internal evaluation purposes only. For all other uses, including academic research, a licence must be acquired (non-commercial (academic) licence, commercial licence).


Download an example sequence (34.4MB, AVI) of "ba ba ba..."

[Homepage] [AVOZES Homepage] [Research]

© Roland Göcke
Last modified: Tue Nov 09 17:25:48 AUS Eastern Daylight Time 2004