Using A* for the parallelization of speech recognition systems

/in , , /by
Using A* for the parallelization of speech recognition systems

Cardinal, Patrick and Boulianne, Gilles and Dumouchel, Pierre

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing – Proceedings 2012

Abstract : The speed of modern processors has remained constant over the last few years but the integration capacity continues to follow Moore’s law and thus, to be scalable, applications must be parallelized. This paper presents results in using the A* search algorithm in a large vocabulary speech recognition parallel system. This algorithm allows better parallelization over the Viterbi algorithm. First experiments with a “unigram approximation” heuristic resulted in approximatively 8.7 times less states being explored compared to our classical Viterbi decoder. The multi-thread implementation of the A* decoder led to a speed-up factor of 3 over its sequential counterpart. © 2012 IEEE.

You might also like
Deep interpretable classification and weakly-supervised segmentation of histology images via max-min uncertainty
Face recognition in video using a What-and-Where fusion neural network
The A* speech recognition system on parallel architectures
Modeling minimum miscibility pressure of pure/impure CO2-crude oil systems using adaptive boosting support vector regression: Application to gas injection processes
Detecting left ventricular impaired relaxation using MR imaging
Annealing based approach to optimize classification system
Combinando caracteristicas complementares em modelos escondidos de markov: uma otimizaçao para o reconhecimento de caracteres manuscritos
Contextual anomaly detection using log-linear tensor factorization