5th European Conference on Speech Communication and Technology

Rhodes, Greece
September 22-25, 1997

Design and Analysis of a German Telephone Speech Database for Phoneme Based Training

Stefan Feldes (1), Bernhard Kaspar (1), Denis Jouvet (2)

(1) Research Group Speech Processing, Deutsche Telekom Berkom, Darmstadt, Germany (2) France Telecom - CNET- LAA/TSS/RCP, Lannion, France

Based on the Sotscheck text corpus, we developped a new corpus that was specifically optimised for training phoneme-based recognition systems. Particular attention was payed on good coverage of phone transitions. Even though the resulting corpus is only slightly enlarged, it shows an increased phonetic coverage while maintaining a good phonetic balance. Results of phonetic statistical analysis and of experiments for training an allophone-based recognizer are reported here.

Full Paper

Bibliographic reference.  Feldes, Stefan / Kaspar, Bernhard / Jouvet, Denis (1997): "Design and analysis of a German telephone speech database for phoneme based training", In EUROSPEECH-1997, 1703-1706.