Navigating the Minefield of MT Beam Search in Cascaded Streaming Speech Translation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Rabatin, Rastislav, Seide, Frank, Chang, Ernie
Format: Preprint
Publié: 2024
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866910528491749376
author Rabatin, Rastislav
Seide, Frank
Chang, Ernie
author_facet Rabatin, Rastislav
Seide, Frank
Chang, Ernie
contents We adapt the well-known beam-search algorithm for machine translation to operate in a cascaded real-time speech translation system. This proved to be more complex than initially anticipated, due to four key challenges: (1) real-time processing of intermediate and final transcriptions with incomplete words from ASR, (2) emitting intermediate and final translations with minimal user perceived latency, (3) handling beam search hypotheses that have unequal length and different model state, and (4) handling sentence boundaries. Previous work in the field of simultaneous machine translation only implemented greedy decoding. We present a beam-search realization that handles all of the above, providing guidance through the minefield of challenges. Our approach increases the BLEU score by 1 point compared to greedy search, reduces the CPU time by up to 40% and character flicker rate by 20+% compared to a baseline heuristic that just retranslates input repeatedly.
format Preprint
id arxiv_https___arxiv_org_abs_2407_11010
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Navigating the Minefield of MT Beam Search in Cascaded Streaming Speech Translation
Rabatin, Rastislav
Seide, Frank
Chang, Ernie
Computation and Language
Machine Learning
Audio and Speech Processing
We adapt the well-known beam-search algorithm for machine translation to operate in a cascaded real-time speech translation system. This proved to be more complex than initially anticipated, due to four key challenges: (1) real-time processing of intermediate and final transcriptions with incomplete words from ASR, (2) emitting intermediate and final translations with minimal user perceived latency, (3) handling beam search hypotheses that have unequal length and different model state, and (4) handling sentence boundaries. Previous work in the field of simultaneous machine translation only implemented greedy decoding. We present a beam-search realization that handles all of the above, providing guidance through the minefield of challenges. Our approach increases the BLEU score by 1 point compared to greedy search, reduces the CPU time by up to 40% and character flicker rate by 20+% compared to a baseline heuristic that just retranslates input repeatedly.
title Navigating the Minefield of MT Beam Search in Cascaded Streaming Speech Translation
topic Computation and Language
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2407.11010