DiscreteSLU: A Large Language Model with Self-Supervised Discrete Speech Units for Spoken Language Understanding

TL;DR

This paper introduces DiscreteSLU, a speech understanding model that uses self-supervised discrete speech units instead of continuous features, improving robustness and instruction-following in spoken language tasks.

Contribution

It proposes a novel approach of converting self-supervised speech encoder outputs into discrete units for better speech understanding in LLMs.

Findings

Robust performance on seen and unseen domains

Effective instruction-following in spoken question answering

Discrete units outperform continuous features in some tasks

Abstract

The integration of pre-trained text-based large language models (LLM) with speech input has enabled instruction-following capabilities for diverse speech tasks. This integration requires the use of a speech encoder, a speech adapter, and an LLM, trained on diverse tasks. We propose the use of discrete speech units (DSU), rather than continuous-valued speech encoder outputs, that are converted to the LLM token embedding space using the speech adapter. We generate DSU using a self-supervised speech encoder followed by k-means clustering. The proposed model shows robust performance on speech inputs from seen/unseen domains and instruction-following capability in spoken question answering. We also explore various types of DSU extracted from different layers of the self-supervised speech encoder, as well as Mel frequency Cepstral Coefficients (MFCC). Our findings suggest that the ASR task and datasets are not crucial in instruction-tuning for spoken question answering tasks.