Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yuan, Junming, Shi, Ying, Li, LanTian, Wang, Dong, Hamdulla, Askar
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2407.06078
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916315889926144
author Yuan, Junming
Shi, Ying
Li, LanTian
Wang, Dong
Hamdulla, Askar
author_facet Yuan, Junming
Shi, Ying
Li, LanTian
Wang, Dong
Hamdulla, Askar
contents Few-shot keyword spotting (KWS) aims to detect unknown keywords with limited training samples. A commonly used approach is the pre-training and fine-tuning framework. While effective in clean conditions, this approach struggles with mixed keyword spotting -- simultaneously detecting multiple keywords blended in an utterance, which is crucial in real-world applications. Previous research has proposed a Mix-Training (MT) approach to solve the problem, however, it has never been tested in the few-shot scenario. In this paper, we investigate the possibility of using MT and other relevant methods to solve the two practical challenges together: few-shot and mixed speech. Experiments conducted on the LibriSpeech and Google Speech Command corpora demonstrate that MT is highly effective on this task when employed in either the pre-training phase or the fine-tuning phase. Moreover, combining SSL-based large-scale pre-training (HuBert) and MT fine-tuning yields very strong results in all the test conditions.
format Preprint
id arxiv_https___arxiv_org_abs_2407_06078
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Few-Shot Keyword Spotting from Mixed Speech
Yuan, Junming
Shi, Ying
Li, LanTian
Wang, Dong
Hamdulla, Askar
Sound
Few-shot keyword spotting (KWS) aims to detect unknown keywords with limited training samples. A commonly used approach is the pre-training and fine-tuning framework. While effective in clean conditions, this approach struggles with mixed keyword spotting -- simultaneously detecting multiple keywords blended in an utterance, which is crucial in real-world applications. Previous research has proposed a Mix-Training (MT) approach to solve the problem, however, it has never been tested in the few-shot scenario. In this paper, we investigate the possibility of using MT and other relevant methods to solve the two practical challenges together: few-shot and mixed speech. Experiments conducted on the LibriSpeech and Google Speech Command corpora demonstrate that MT is highly effective on this task when employed in either the pre-training phase or the fine-tuning phase. Moreover, combining SSL-based large-scale pre-training (HuBert) and MT fine-tuning yields very strong results in all the test conditions.
title Few-Shot Keyword Spotting from Mixed Speech
topic Sound
url https://arxiv.org/abs/2407.06078