XLSR-Transducer: Streaming ASR for Self-Supervised Pretrained Models

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Kumar, Shashi, Madikeri, Srikanth, Zuluaga-Gomez, Juan, Villatoro-Tello, Esaú, Thorbecke, Iuliia, Motlicek, Petr, E, Manjunath K, Ganapathiraju, Aravind
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910640250028032
author Kumar, Shashi
Madikeri, Srikanth
Zuluaga-Gomez, Juan
Villatoro-Tello, Esaú
Thorbecke, Iuliia
Motlicek, Petr
E, Manjunath K
Ganapathiraju, Aravind
author_facet Kumar, Shashi
Madikeri, Srikanth
Zuluaga-Gomez, Juan
Villatoro-Tello, Esaú
Thorbecke, Iuliia
Motlicek, Petr
E, Manjunath K
Ganapathiraju, Aravind
contents Self-supervised pretrained models exhibit competitive performance in automatic speech recognition on finetuning, even with limited in-domain supervised data. However, popular pretrained models are not suitable for streaming ASR because they are trained with full attention context. In this paper, we introduce XLSR-Transducer, where the XLSR-53 model is used as encoder in transducer setup. Our experiments on the AMI dataset reveal that the XLSR-Transducer achieves 4% absolute WER improvement over Whisper large-v2 and 8% over a Zipformer transducer model trained from scratch. To enable streaming capabilities, we investigate different attention masking patterns in the self-attention computation of transformer layers within the XLSR-53 model. We validate XLSR-Transducer on AMI and 5 languages from CommonVoice under low-resource scenarios. Finally, with the introduction of attention sinks, we reduce the left context by half while achieving a relative 12% improvement in WER.
format Preprint
id arxiv_https___arxiv_org_abs_2407_04439
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle XLSR-Transducer: Streaming ASR for Self-Supervised Pretrained Models
Kumar, Shashi
Madikeri, Srikanth
Zuluaga-Gomez, Juan
Villatoro-Tello, Esaú
Thorbecke, Iuliia
Motlicek, Petr
E, Manjunath K
Ganapathiraju, Aravind
Audio and Speech Processing
Self-supervised pretrained models exhibit competitive performance in automatic speech recognition on finetuning, even with limited in-domain supervised data. However, popular pretrained models are not suitable for streaming ASR because they are trained with full attention context. In this paper, we introduce XLSR-Transducer, where the XLSR-53 model is used as encoder in transducer setup. Our experiments on the AMI dataset reveal that the XLSR-Transducer achieves 4% absolute WER improvement over Whisper large-v2 and 8% over a Zipformer transducer model trained from scratch. To enable streaming capabilities, we investigate different attention masking patterns in the self-attention computation of transformer layers within the XLSR-53 model. We validate XLSR-Transducer on AMI and 5 languages from CommonVoice under low-resource scenarios. Finally, with the introduction of attention sinks, we reduce the left context by half while achieving a relative 12% improvement in WER.
title XLSR-Transducer: Streaming ASR for Self-Supervised Pretrained Models
topic Audio and Speech Processing
url https://arxiv.org/abs/2407.04439