An Exploration of Mamba for Speech Self-Supervised Models

TL;DR

This paper investigates Mamba-based HuBERT models for speech self-supervised learning, demonstrating their efficiency and superior performance in long-context ASR, streaming, and speech representation tasks.

Contribution

It introduces Mamba-based HuBERT models as efficient alternatives to Transformer SSL architectures, with improved performance and lower computational costs.

Findings

Mamba-based models outperform Transformer models in streaming ASR.

They produce higher-quality quantized speech representations.

Models show competitive results on SUPERB benchmarks.

Abstract

While Mamba has demonstrated strong performance in language modeling, its potential as a speech self-supervised learning (SSL) model remains underexplored, with prior studies limited to isolated tasks. To address this, we explore Mamba-based HuBERT models as alternatives to Transformer-based SSL architectures. Leveraging the linear-time Selective State Space, these models enable fine-tuning on long-context ASR with significantly lower compute. Moreover, they show superior performance when fine-tuned for streaming ASR. Beyond fine-tuning, these models show competitive performance on SUPERB probing benchmarks, particularly in causal settings. Our analysis shows that they yield higher-quality quantized representations and capture speaker-related features more distinctly than Transformer-based models. These findings highlight Mamba-based SSL as a promising and complementary direction for long-sequence modeling, real-time speech modeling, and speech unit extraction. The codebase is available at https://github.com/hckuo145/Mamba-based-HuBERT.