Title :
Discriminative resolution enhancement in acoustic modelling
Author :
Duchateau, Jacques ; Demuynck, Kris ; Wambacq, Patrick
Author_Institution :
ESAT, Katholieke Univ., Leuven, Heverlee, Belgium
Abstract :
The accuracy of the acoustic models in large vocabulary recognition systems can be improved by increasing the resolution in the acoustic feature space. This can be obtained by increasing the number of Gaussian densities in the models by splitting of the Gaussians. This paper proposes a novel algorithm for this splitting operation. It is based on the phonetic decision tree used for the state tying in context dependent modelling. The advantage of the method is that it improves the capability of the acoustic models to discriminate between the different tied states. The proposed splitting algorithm was evaluated on the Wall Street Journal recognition task. Comparison with a commonly used splitting algorithm clearly shows that our method can provide smaller (thus faster) acoustic models and results in lower error rates
Keywords :
Gaussian processes; decision trees; speech enhancement; speech recognition; Gaussian densities; acoustic feature space; acoustic modelling; context dependent modelling; discriminative resolution enhancement; error rates; large vocabulary recognition systems; phonetic decision tree; state tying; Context modeling; Decision trees; Error analysis; Mutual information; Speech recognition; Tail; Training data; Vocabulary;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 2000. ICASSP '00. Proceedings. 2000 IEEE International Conference on
Conference_Location :
Istanbul
Print_ISBN :
0-7803-6293-4
DOI :
10.1109/ICASSP.2000.861801