Parallelizing WFST speech decoders
Lists
Only you see this
Abstract
The performance-intensive part of a large-vocabulary continuous speech-recognition system is the Viterbi computation that determines the sequence of words that are most likely to generate the acoustic-state scores extracted from an input utterance. This paper presents an efficient parallel algorithm for Viterbi. The key idea is to partition the per-frame computation among threads to minimize inter-thread communication despite traversing a large irregular acoustic and language model graphs. Together with a per-thread beam search, load balancing language-model lookups, and memory optimizations, we achieve a 6.67x speedup over an highly-optimized production-quality WFST-based speech decoder. On a 200,000 word vocabulary and a 59 million ngram model, our decoder runs at 0.27x real time while achieving a word-error rate of 14.81% on 6214 labeled utterances from Voice Search data.
BibTeX
@inproceedings{mendis2016parallelizing,
author = {Charith Mendis and Jasha Droppo and Saeed Maleki and Madanlal Musuvathi and Todd Mytkowicz and Geoffrey Zweig},
title = {{Parallelizing WFST speech decoders}},
booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
year = {2016},
doi = {10.1109/icassp.2016.7472694},
}Loading related content...