CMU's IWSLT 2024 Simultaneous Speech Translation System

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Xu, Xi, Ouyang, Siqi, Yan, Brian, Fernandes, Patrick, Chen, William, Li, Lei, Neubig, Graham, Watanabe, Shinji
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914912858537984
author Xu, Xi
Ouyang, Siqi
Yan, Brian
Fernandes, Patrick
Chen, William
Li, Lei
Neubig, Graham
Watanabe, Shinji
author_facet Xu, Xi
Ouyang, Siqi
Yan, Brian
Fernandes, Patrick
Chen, William
Li, Lei
Neubig, Graham
Watanabe, Shinji
contents This paper describes CMU's submission to the IWSLT 2024 Simultaneous Speech Translation (SST) task for translating English speech to German text in a streaming manner. Our end-to-end speech-to-text (ST) system integrates the WavLM speech encoder, a modality adapter, and the Llama2-7B-Base model as the decoder. We employ a two-stage training approach: initially, we align the representations of speech and text, followed by full fine-tuning. Both stages are trained on MuST-c v2 data with cross-entropy loss. We adapt our offline ST model for SST using a simple fixed hold-n policy. Experiments show that our model obtains an offline BLEU score of 31.1 and a BLEU score of 29.5 under 2 seconds latency on the MuST-C-v2 tst-COMMON.
format Preprint
id arxiv_https___arxiv_org_abs_2408_07452
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle CMU's IWSLT 2024 Simultaneous Speech Translation System
Xu, Xi
Ouyang, Siqi
Yan, Brian
Fernandes, Patrick
Chen, William
Li, Lei
Neubig, Graham
Watanabe, Shinji
Computation and Language
Artificial Intelligence
This paper describes CMU's submission to the IWSLT 2024 Simultaneous Speech Translation (SST) task for translating English speech to German text in a streaming manner. Our end-to-end speech-to-text (ST) system integrates the WavLM speech encoder, a modality adapter, and the Llama2-7B-Base model as the decoder. We employ a two-stage training approach: initially, we align the representations of speech and text, followed by full fine-tuning. Both stages are trained on MuST-c v2 data with cross-entropy loss. We adapt our offline ST model for SST using a simple fixed hold-n policy. Experiments show that our model obtains an offline BLEU score of 31.1 and a BLEU score of 29.5 under 2 seconds latency on the MuST-C-v2 tst-COMMON.
title CMU's IWSLT 2024 Simultaneous Speech Translation System
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2408.07452