Joint speech and overlap detection: a benchmark over multiple audio setup and speech domains
This work addresses the need for robust pre-processing in speaker diarization, but it is incremental as it builds on existing joint training methods by extending them to more diverse conditions.
The paper tackles the problem of voice activity and overlapped speech detection by proposing a joint training benchmark across multiple audio setups and speech domains, achieving state-of-the-art F1-scores with a system that reduces training costs compared to dedicated models.
Voice activity and overlapped speech detection (respectively VAD and OSD) are key pre-processing tasks for speaker diarization. The final segmentation performance highly relies on the robustness of these sub-tasks. Recent studies have shown VAD and OSD can be trained jointly using a multi-class classification model. However, these works are often restricted to a specific speech domain, lacking information about the generalization capacities of the systems. This paper proposes a complete and new benchmark of different VAD and OSD models, on multiple audio setups (single/multi-channel) and speech domains (e.g. media, meeting...). Our 2/3-class systems, which combine a Temporal Convolutional Network with speech representations adapted to the setup, outperform state-of-the-art results. We show that the joint training of these two tasks offers similar performances in terms of F1-score to two dedicated VAD and OSD systems while reducing the training cost. This unique architecture can also be used for single and multichannel speech processing.