On-device Streaming Discrete Speech Units

Kwanghee Choi, Masao Someki, Emma Strubell, Shinji Watanabe

Keywords: discrete speech units, streaming, on-device, speech recognition

TL;DR: We make discrete speech units (from self-supervised speech models) more lightweight by reducing the number of layers and attention window size.

Presentation Preference: Yes

Abstract: Discrete speech units (DSUs) are derived from clustering the features of self-supervised speech models (S3Ms). DSUs offer significant advantages for on-device streaming speech applications due to their rich phonetic information, high transmission efficiency, and seamless integration with large language models. However, conventional DSU-based approaches are impractical as they require full-length speech input and computationally expensive S3Ms. In this work, we reduce both the attention window and the model size while preserving the effectiveness of DSUs. Our results demonstrate that we can reduce floating-point operations (FLOPs) by 50% with only a relative increase of 6.5% in character error rate (CER) on the ML-SUPERB 1h dataset. These findings highlight the potential of DSUs for real-time speech processing in resource-constrained environments.

Submission Number: 5