Efficient and Robust Long-Form Speech Recognition with Hybrid H3-Conformer

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Honda, Tomoki, Sakai, Shinsuke, Kawahara, Tatsuya
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917795668688896
author Honda, Tomoki
Sakai, Shinsuke
Kawahara, Tatsuya
author_facet Honda, Tomoki
Sakai, Shinsuke
Kawahara, Tatsuya
contents Recently, Conformer has achieved state-of-the-art performance in many speech recognition tasks. However, the Transformer-based models show significant deterioration for long-form speech, such as lectures, because the self-attention mechanism becomes unreliable with the computation of the square order of the input length. To solve the problem, we incorporate a kind of state-space model, Hungry Hungry Hippos (H3), to replace or complement the multi-head self-attention (MHSA). H3 allows for efficient modeling of long-form sequences with a linear-order computation. In experiments using two datasets of CSJ and LibriSpeech, our proposed H3-Conformer model performs efficient and robust recognition of long-form speech. Moreover, we propose a hybrid of H3 and MHSA and show that using H3 in higher layers and MHSA in lower layers provides significant improvement in online recognition. We also investigate a parallel use of H3 and MHSA in all layers, resulting in the best performance.
format Preprint
id arxiv_https___arxiv_org_abs_2410_04159
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Efficient and Robust Long-Form Speech Recognition with Hybrid H3-Conformer
Honda, Tomoki
Sakai, Shinsuke
Kawahara, Tatsuya
Sound
Audio and Speech Processing
Recently, Conformer has achieved state-of-the-art performance in many speech recognition tasks. However, the Transformer-based models show significant deterioration for long-form speech, such as lectures, because the self-attention mechanism becomes unreliable with the computation of the square order of the input length. To solve the problem, we incorporate a kind of state-space model, Hungry Hungry Hippos (H3), to replace or complement the multi-head self-attention (MHSA). H3 allows for efficient modeling of long-form sequences with a linear-order computation. In experiments using two datasets of CSJ and LibriSpeech, our proposed H3-Conformer model performs efficient and robust recognition of long-form speech. Moreover, we propose a hybrid of H3 and MHSA and show that using H3 in higher layers and MHSA in lower layers provides significant improvement in online recognition. We also investigate a parallel use of H3 and MHSA in all layers, resulting in the best performance.
title Efficient and Robust Long-Form Speech Recognition with Hybrid H3-Conformer
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2410.04159