MAC: A unified framework boosting low resource automatic speech recognition

TL;DR

This paper introduces MAC, a unified Bayesian sampling-based framework that enhances low-resource automatic speech recognition by integrating a novel concatenative synthesis TTS system, significantly reducing error rates across multiple languages.

Contribution

The paper presents MAC, a novel low-resource ASR framework that combines Bayesian sampling with concatenative synthesis TTS, improving recognition accuracy and handling diverse languages and scenes.

Findings

MAC reduces CER by over 15% in multiple language ASR tasks.

MAC outperforms wav2vec2 on Cantonese, Taiwanese, and Japanese datasets.

Achieves 10.9% CER on Cantonese ASR, a 30% relative improvement.

Abstract

We propose a unified framework for low resource automatic speech recognition tasks named meta audio concatenation (MAC). It is easy to implement and can be carried out in extremely low resource environments. Mathematically, we give a clear description of MAC framework from the perspective of bayesian sampling. In this framework, we leverage a novel concatenative synthesis text-to-speech system to boost the low resource ASR task. By the concatenative synthesis text-to-speech system, we can integrate language pronunciation rules and adjust the TTS process. Furthermore, we propose a broad notion of meta audio set to meet the modeling needs of different languages and different scenes when using the system. Extensive experiments have demonstrated the great effectiveness of MAC on low resource ASR tasks. For CTC greedy search, CTC prefix, attention, and attention rescoring decode mode in Cantonese ASR task, Taiwanese ASR task, and Japanese ASR task the MAC method can reduce the CER by more than 15\%. Furthermore, in the ASR task, MAC beats wav2vec2 (with fine-tuning) on common voice datasets of Cantonese and gets really competitive results on common voice datasets of Taiwanese and Japanese. Among them, it is worth mentioning that we achieve a \textbf{10.9\%} character error rate (CER) on the common voice Cantonese ASR task, bringing about \textbf{30\%} relative improvement compared to the wav2vec2 (with fine-tuning).