Loïc Barrault
Laboratoire Informatique d'Avignon
Session JEP poster P3 Mardi 10 Juin - 14h00 16h00
-
papier 1603
Combinaison de différents jeux de paramètres acoustiques pour la reconnaissance de la parole
- Loïc Barrault ( Laboratoire Informatique d'Avignon)
- Driss Matrouf ( Laboratoire Informatique d'Avignon)
- Georges Linarès ( Laboratoire Informatique d'Avignon)
- Renato De-Mori ( Laboratoire Informatique d'Avignon)
- Résumé : With the purpose of improving Automatic Speech Recognition (ASR) systems performance, many different approaches on combining them have been largely studied. In this paper, a combination of state a posteriori probabilities given by different feature sets is proposed. In order to perform a coherent combination of state posterior probabilities, the acoustic models trained on different feature sets must have the same topology (i.e. same set of states). For this purpose, a fast and efficient twin model training protocol is proposed. Two different strategies for combining probabilities are presented : the linear and the log linear interpolation. By using log linear interpolation, a relative Word Error Rate (WER) reduction of about 15% and 14% have been observed respectively on MEDIA and ESTER corpora.
- article