10 hours data is all you need

TL;DR

This paper introduces CAMP, a method for generating pseudo Mandarin speech data from characters, and META-AUDIO, a scalable database building approach, enabling effective speech recognition with only 10 hours of real data.

Contribution

The paper presents CAMP and META-AUDIO, novel techniques for data augmentation and database construction that improve Mandarin speech recognition with limited data.

Findings

Achieved 11.07 CER on AISHELL-1 with 10 hours of real and pseudo data.

Achieved 8.26 CER on AIDATATANG with limited data and pseudo data.

Demonstrated effectiveness of pseudo data in reducing data requirements.

Abstract

We propose a novel procedure to generate pseudo mandarin speech data named as CAMP (character audio mix up), which aims at generating audio from a character scale. We also raise a method for building a mandarin character scale audio database adaptive to CAMP named as META-AUDIO, which makes full use of audio data and can greatly increase the data diversity of the database. Experiments show that our CAMP method is simple and quite effective. For example, we train models with 10 hours of audio data in AISHELL-1 and pseudo audio data generated by CAMP, and achieve a competitive 11.07 character error rate (CER). Besides, we also perform training with only 10 hours of audio data in AIDATATANG dataset and pseudo audio data generated by CAMP, which again achieves a competitive 8.26 CER.