Multi-scenario deep learning for multi-speaker source separation

TL;DR

This paper demonstrates that training a single deep learning model on multiple multi-speaker scenarios can match the performance of models trained on specific scenarios, highlighting the importance of diverse data in source separation tasks.

Contribution

It introduces a multi-scenario training approach for deep learning models in multi-speaker source separation, showing its effectiveness across different scenarios.

Findings

Data from one scenario improves performance in another.

Single multi-scenario model matches scenario-specific models.

Diverse training data enhances model generalization.

Abstract

Research in deep learning for multi-speaker source separation has received a boost in the last years. However, most studies are restricted to mixtures of a specific number of speakers, called a specific scenario. While some works included experiments for different scenarios, research towards combining data of different scenarios or creating a single model for multiple scenarios have been very rare. In this work it is shown that data of a specific scenario is relevant for solving another scenario. Furthermore, it is concluded that a single model, trained on different scenarios is capable of matching performance of scenario specific models.