Ask2Mask: Guided Data Selection for Masked Speech Modeling

doi:10.1109/JSTSP.2022.3186162

2022-10-01

Murali Karthick Baskar *

Bhuvana Ramabhadran

Yu Zhang

DOI：10.1109/JSTSP.2022.3186162

原文链接

原文求助

摘要

En 中文

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 in two different ways: 1) A fine-grained data selection is performed by masking over the highly confident input frames as chosen by the scorer. This allows the model to learn meaningful representations. 2) ATM is further extended to focus at utterance-level by weighting the final MSM loss with the utterance-level confidence score. We conduct fine-tuning experiments on two well-benchmarked corpora: LibriSpeech (matching the pre-training data) and Commonvoice, TED-LIUM, 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 (up to 11.6% relative over published results and upto 4.46% relative over our internal baseline) while still yielding modest improvements under matched conditions.

Keyword:

Data models

Training

Computational modeling

Bit error rate

Context modeling

Training data

Task analysis

Self-supervision

Wav2vec2

Data selection

Domain mismatch