Dual-path Mamba: Short and Long-term Bidirectional Selective Structured State Space Models for Speech Separation
Authors:
Top Cited Papers
- 6 April 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Abstract
Transformers have been the most successful architecture for various speech modeling tasks, including speech separation. However, the self-attention mechanism in transformers with quadratic complexity is inefficient in computation and memory. Recent models incorporate new layers and modules along with transformers for better performance but also introduce extra model complexity. In this work, we replace transformers with Mamba, a selective state space model, for speech separation. We propose dual-path Mamba, which models short-term and long-term forward and backward dependency of speech signals using selective state spaces. Our experimental results on the WSJ0-2mix data show that our dual-path Mamba models of comparably smaller sizes outperform state-of-the-art RNN model DPRNN, CNN model WaveSplit, and transformer model Sepformer. Our largest model reaches an SI-SNRi of 23.4 dB. Code available: https://github.com/xi-j/Mamba-TasNetKeywords
All Related Versions
Funding Information
- Health
This publication has 25 references indexed in Scilit:
- Motion Mamba: Efficient and Long Sequence Motion Generation2024
- MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation2024
- On Time Domain Conformer Models for Monaural Speech Separation in Noisy Reverberant Acoustic Environments2023
- Compute and Memory Efficient Universal Sound Source SeparationJournal of Signal Processing Systems, 2021
- Attention Is All You Need In Speech Separation2021
- Dual-Path RNN: Efficient Long Sequence Modeling for Time-Domain Single-Channel Speech Separation2020
- SDR – Half-baked or Well Done?2019
- End-To-End Source Separation With Adaptive Front-Ends2018
- TaSNet: Time-Domain Audio Separation Network for Real-Time, Single-Channel Speech Separation2018
- Deep clustering: Discriminative embeddings for segmentation and separation2016