Schrödinger Bridge Mamba for One-Step Speech Enhancement
This work addresses speech enhancement for applications requiring real-time processing, presenting a novel framework that is potentially applicable to broader generative tasks.
The authors tackled speech enhancement by proposing Schrödinger Bridge Mamba (SBM), a training-inference framework that combines Schrödinger Bridge training with Mamba's selective state-space model, achieving state-of-the-art performance with only 1-step inference on joint denoising and dereverberation tasks across four benchmark datasets and the best real-time factor.
We propose Schrödinger Bridge Mamba (SBM), a new concept of training-inference framework motivated by the inherent compatibility between Schrödinger Bridge (SB) training paradigm and selective state-space model Mamba. We exemplify the concept of SBM with an implementation for generative speech enhancement. Experiments on a joint denoising and dereverberation task using four benchmark datasets demonstrate that SBM, with only 1-step inference, outperforms strong baselines with 1-step or iterative inference and achieves the best real-time factor (RTF). Beyond speech enhancement, we discuss the integration of SB paradigm and selective state-space model architecture based on their underlying alignment, which indicates a promising direction for exploring new deep generative models potentially applicable to a broad range of generative tasks. Demo page: https://sbmse.github.io