Robustness of Speech Separation Models for Similar-pitch Speakers
This addresses a problem for speech recognition systems in multi-speaker environments, but it is incremental as it extends prior analysis to newer models.
This paper investigates the robustness of state-of-the-art Neural Network models for single-channel speech separation in scenarios with minimal pitch differences between speakers, finding that modern models reduce performance gaps under matched conditions but still show substantial gaps under mismatched conditions.
Single-channel speech separation is a crucial task for enhancing speech recognition systems in multi-speaker environments. This paper investigates the robustness of state-of-the-art Neural Network models in scenarios where the pitch differences between speakers are minimal. Building on earlier findings by Ditter and Gerkmann, which identified a significant performance drop for the 2018 Chimera++ under similar-pitch conditions, our study extends the analysis to more recent and sophisticated Neural Network models. Our experiments reveal that modern models have substantially reduced the performance gap for matched training and testing conditions. However, a substantial performance gap persists under mismatched conditions, with models performing well for large pitch differences but showing worse performance if the speakers' pitches are similar. These findings motivate further research into the generalizability of speech separation models to similar-pitch speakers and unseen data.