next up previous
Next: RDS: Reverse directory service Up: Some of the completed Previous: POSSY/TTS'99: Polyglot speech synthesis

 

Polyphone: The speech data collection project

In the framework of this project speech samples of 5000 speakers from the German speaking part of Switzerland have been collected via telephone. The aim was to achieve a good coverage of the variability in terms of voice type, speaker environment, and transmission channel characteristics.

The recorded speech includes 46 items, of which some are spontaneous and some are read. The majority of the items were spoken in standard German and a few items were uttered in dialect.

Each of the recruited speakers got an individual prompt sheet. These sheets had been designed to reach a good phonetic coverage per speaker and for all the speakers. The speakers had to phone to an automatic recording equipment and follow the recording dialogue. The recorded speech was manually annotated in textual form (further information in [Hub98a]).

Supported by: The project was fully financed by Swisscom.

In collaboration with: IDIAP, Swisscom and the Link Marketing Institute in Lausanne.




Last updated: Thu Oct 27 14:58:12 CEST 2016 by: Beat Pfister