next up previous contents
Next: Conclusion Up: Task 7.3: Demonstrations Previous: Task 7.3: Objective

Task 7.3: Status

The AbbotDemo is now including our baseline speech recognition system in French, proposing a 20,000 word vocabulary continuous speech recognition interface. So far, the system is using feed-forward MLP, context-independent phonemes and trigrams extracted from French newspapers.

An isolated word recognizer running in a Windows environment, with vocabulary designed on the spot, has been implemented.

Christophe Ris