Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Simultaneous perturbation stochastic approximation for automatic speech recognition

: Stein, D.; Schwenninger, J.; Stadtschnitzer, M.

Bimbot, F. ; International Speech Communication Association -ISCA-:
14th Annual Conference of the International Speech Communication Association, INTERSPEECH 2013 : Lyon, France, August 25-29, 2013
Lyon: ISCA, 2013
International Speech Communication Association (Annual Conference INTERSPEECH) <14, 2013, Lyon>
Fraunhofer IAIS ()

While both the acoustic model and the language model in automatic speech recognition are typically well-trained on the target domain, the free parameters of the decoder itself are often set manually. In this paper, we investigate in how far a stochastic approximation algorithm can be employed to automatically determine the best parameters, especially if additional timeconstraints are given on unknown machine architectures. We offer our findings on the German Difficult Speech Corpus, and present significant improvements over both the spontaneous and planned clean speech task. Copyright