Reducing Domain mismatch in Self-supervised speech pre-training

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F22%3APU146109" target="_blank" >RIV/00216305:26230/22:PU146109 - isvavai.cz</a>

Result on the web

<a href="https://www.isca-speech.org/archive/pdfs/interspeech_2022/baskar22_interspeech.pdf" target="_blank" >https://www.isca-speech.org/archive/pdfs/interspeech_2022/baskar22_interspeech.pdf</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.21437/Interspeech.2022-736" target="_blank" >10.21437/Interspeech.2022-736</a>

Result language

angličtina

Original language name

Reducing Domain mismatch in Self-supervised speech pre-training

Original language description

Masked speech modeling (MSM) methods such as wav2vec2 or w2v-BERT learn representations over speech frames which are randomly masked within an utterance. While these methods improve performance of Automatic Speech Recognition (ASR) systems, they have one major limitation. They treat all unsupervised speech samples with equal weight, which hinders learning as not all samples have relevant information to learn meaningful representations. In this work, we address this limitation. We propose ask2mask (ATM), a novel approach to focus on specific samples during MSM pre-training. ATM employs an external ASR model or scorer to weight unsupervised input samples by performing a fine-grained data selection. ATM performs masking over the highly confident input frames as chosen by the scorer. This allows the model to learn meaningful representations. We conduct fine-tuning experiments on two well-benchmarked corpora: LibriSpeech (matching the pre-training data) and, AMI and CHiME-6 (not matching the pre-training data). The results substantiate the efficacy of ATM on significantly improving the recognition performance under mismatched conditions while still yielding modest improvements under matched conditions.

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

—

Continuities

S - Specificky vyzkum na vysokych skolach

Publication year

2022

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection

Proceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH

ISBN

—

ISSN

1990-9772

e-ISSN

—

Number of pages

5

Pages from-to

3028-3032

Publisher name

International Speech Communication Association

Place of publication

Incheon

Event location

Incheon Korea

Event date

Sep 18, 2022

Type of event by nationality

WRD - Celosvětová akce

UT code for WoS article

—