AS-70: A Mandarin stuttered speech dataset for automatic speech recognition and stuttering event detection

TL;DR

This paper introduces AS-70, the first large publicly available Mandarin stuttered speech dataset, and demonstrates its usefulness in improving speech recognition and stuttering event detection models.

Contribution

It provides the first large Mandarin stuttered speech dataset and baseline systems, enabling better ASR and stuttering detection for atypical speech.

Findings

Significant improvements in ASR accuracy with the dataset

Enhanced stuttering event detection performance

Increased inclusivity of speech models for atypical speech

Abstract

The rapid advancements in speech technologies over the past two decades have led to human-level performance in tasks like automatic speech recognition (ASR) for fluent speech. However, the efficacy of these models diminishes when applied to atypical speech, such as stuttering. This paper introduces AS-70, the first publicly available Mandarin stuttered speech dataset, which stands out as the largest dataset in its category. Encompassing conversational and voice command reading speech, AS-70 includes verbatim manual transcription, rendering it suitable for various speech-related tasks. Furthermore, baseline systems are established, and experimental results are presented for ASR and stuttering event detection (SED) tasks. By incorporating this dataset into the model fine-tuning, significant improvements in the state-of-the-art ASR models, e.g., Whisper and Hubert, are observed, enhancing their inclusivity in addressing stuttered speech.