Efficient Long-Form Speech Recognition for General Speech In-Context Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Yen, Hao, Ling, Shaoshi, Ye, Guoli
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913522713100288
author Yen, Hao
Ling, Shaoshi
Ye, Guoli
author_facet Yen, Hao
Ling, Shaoshi
Ye, Guoli
contents We propose a novel approach to end-to-end automatic speech recognition (ASR) to achieve efficient speech in-context learning (SICL) for (i) long-form speech decoding, (ii) test-time speaker adaptation, and (iii) test-time contextual biasing. Specifically, we introduce an attention-based encoder-decoder (AED) model with SICL capability (referred to as SICL-AED), where the decoder utilizes an utterance-level cross-attention to integrate information from the encoder's output efficiently, and a document-level self-attention to learn contextual information. Evaluated on the benchmark TEDLIUM3 dataset, SICL-AED achieves an 8.64% relative word error rate (WER) reduction compared to a baseline utterance-level AED model by leveraging previously decoded outputs as in-context examples. It also demonstrates comparable performance to conventional long-form AED systems with significantly reduced runtime and memory complexity. Additionally, we introduce an in-context fine-tuning (ICFT) technique that further enhances SICL effectiveness during inference. Experiments on speaker adaptation and contextual biasing highlight the general speech in-context learning capabilities of our system, achieving effective results with provided contexts. Without specific fine-tuning, SICL-AED matches the performance of supervised AED baselines for speaker adaptation and improves entity recall by 64% for contextual biasing task.
format Preprint
id arxiv_https___arxiv_org_abs_2409_19757
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Efficient Long-Form Speech Recognition for General Speech In-Context Learning
Yen, Hao
Ling, Shaoshi
Ye, Guoli
Audio and Speech Processing
Sound
We propose a novel approach to end-to-end automatic speech recognition (ASR) to achieve efficient speech in-context learning (SICL) for (i) long-form speech decoding, (ii) test-time speaker adaptation, and (iii) test-time contextual biasing. Specifically, we introduce an attention-based encoder-decoder (AED) model with SICL capability (referred to as SICL-AED), where the decoder utilizes an utterance-level cross-attention to integrate information from the encoder's output efficiently, and a document-level self-attention to learn contextual information. Evaluated on the benchmark TEDLIUM3 dataset, SICL-AED achieves an 8.64% relative word error rate (WER) reduction compared to a baseline utterance-level AED model by leveraging previously decoded outputs as in-context examples. It also demonstrates comparable performance to conventional long-form AED systems with significantly reduced runtime and memory complexity. Additionally, we introduce an in-context fine-tuning (ICFT) technique that further enhances SICL effectiveness during inference. Experiments on speaker adaptation and contextual biasing highlight the general speech in-context learning capabilities of our system, achieving effective results with provided contexts. Without specific fine-tuning, SICL-AED matches the performance of supervised AED baselines for speaker adaptation and improves entity recall by 64% for contextual biasing task.
title Efficient Long-Form Speech Recognition for General Speech In-Context Learning
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2409.19757