Multi-Microphone Speaker Separation by Spatial Regions
This work addresses source separation in reverberant environments for audio processing applications, but it is incremental as it modifies an existing state-of-the-art network.
The paper tackles the problem of separating speakers from multi-microphone recordings based on pre-defined spatial regions, achieving a 1.5 dB improvement in scale-invariant signal-to-distortion ratio over a baseline network.
We consider the task of region-based source separation of reverberant multi-microphone recordings. We assume pre-defined spatial regions with a single active source per region. The objective is to estimate the signals from the individual spatial regions as captured by a reference microphone while retaining a correspondence between signals and spatial regions. We propose a data-driven approach using a modified version of a state-of-the-art network, where different layers model spatial and spectro-temporal information. The network is trained to enforce a fixed mapping of regions to network outputs. Using speech from LibriMix, we construct a data set specifically designed to contain the region information. Additionally, we train the network with permutation invariant training. We show that both training methods result in a fixed mapping of regions to network outputs, achieve comparable performance, and that the networks exploit spatial information. The proposed network outperforms a baseline network by 1.5 dB in scale-invariant signal-to-distortion ratio.