Salvato in:
Dettagli Bibliografici
Autori principali: Deng, Keqi, Guo, Jinxi, Ma, Yingyi, Moritz, Niko, Woodland, Philip C., Kalinli, Ozlem, Seltzer, Mike
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2412.16464
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916537842008064
author Deng, Keqi
Guo, Jinxi
Ma, Yingyi
Moritz, Niko
Woodland, Philip C.
Kalinli, Ozlem
Seltzer, Mike
author_facet Deng, Keqi
Guo, Jinxi
Ma, Yingyi
Moritz, Niko
Woodland, Philip C.
Kalinli, Ozlem
Seltzer, Mike
contents While large language models (LLMs) have been applied to automatic speech recognition (ASR), the task of making the model streamable remains a challenge. This paper proposes a novel model architecture, Transducer-Llama, that integrates LLMs into a Factorized Transducer (FT) model, naturally enabling streaming capabilities. Furthermore, given that the large vocabulary of LLMs can cause data sparsity issue and increased training costs for spoken language systems, this paper introduces an efficient vocabulary adaptation technique to align LLMs with speech system vocabularies. The results show that directly optimizing the FT model with a strong pre-trained LLM-based predictor using the RNN-T loss yields some but limited improvements over a smaller pre-trained LM predictor. Therefore, this paper proposes a weak-to-strong LM swap strategy, using a weak LM predictor during RNN-T loss training and then replacing it with a strong LLM. After LM replacement, the minimum word error rate (MWER) loss is employed to finetune the integration of the LLM predictor with the Transducer-Llama model. Experiments on the LibriSpeech and large-scale multi-lingual LibriSpeech corpora show that the proposed streaming Transducer-Llama approach gave a 17% relative WER reduction (WERR) over a strong FT baseline and a 32% WERR over an RNN-T baseline.
format Preprint
id arxiv_https___arxiv_org_abs_2412_16464
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Transducer-Llama: Integrating LLMs into Streamable Transducer-based Speech Recognition
Deng, Keqi
Guo, Jinxi
Ma, Yingyi
Moritz, Niko
Woodland, Philip C.
Kalinli, Ozlem
Seltzer, Mike
Computation and Language
Audio and Speech Processing
While large language models (LLMs) have been applied to automatic speech recognition (ASR), the task of making the model streamable remains a challenge. This paper proposes a novel model architecture, Transducer-Llama, that integrates LLMs into a Factorized Transducer (FT) model, naturally enabling streaming capabilities. Furthermore, given that the large vocabulary of LLMs can cause data sparsity issue and increased training costs for spoken language systems, this paper introduces an efficient vocabulary adaptation technique to align LLMs with speech system vocabularies. The results show that directly optimizing the FT model with a strong pre-trained LLM-based predictor using the RNN-T loss yields some but limited improvements over a smaller pre-trained LM predictor. Therefore, this paper proposes a weak-to-strong LM swap strategy, using a weak LM predictor during RNN-T loss training and then replacing it with a strong LLM. After LM replacement, the minimum word error rate (MWER) loss is employed to finetune the integration of the LLM predictor with the Transducer-Llama model. Experiments on the LibriSpeech and large-scale multi-lingual LibriSpeech corpora show that the proposed streaming Transducer-Llama approach gave a 17% relative WER reduction (WERR) over a strong FT baseline and a 32% WERR over an RNN-T baseline.
title Transducer-Llama: Integrating LLMs into Streamable Transducer-based Speech Recognition
topic Computation and Language
Audio and Speech Processing
url https://arxiv.org/abs/2412.16464