English

Multi-resolution location-based training for multi-channel continuous speech separation

Audio and Speech Processing 2023-01-18 v1 Sound

Abstract

The performance of automatic speech recognition (ASR) systems severely degrades when multi-talker speech overlap occurs. In meeting environments, speech separation is typically performed to improve the robustness of ASR systems. Recently, location-based training (LBT) was proposed as a new training criterion for multi-channel talker-independent speaker separation. Assuming fixed array geometry, LBT outperforms widely-used permutation-invariant training in fully overlapped utterances and matched reverberant conditions. This paper extends LBT to conversational multi-channel speaker separation. We introduce multi-resolution LBT to estimate the complex spectrograms from low to high time and frequency resolutions. With multi-resolution LBT, convolutional kernels are assigned consistently based on speaker locations in physical space. Evaluation results show that multi-resolution LBT consistently outperforms other competitive methods on the recorded LibriCSS corpus.

Keywords

Cite

@article{arxiv.2301.06458,
  title  = {Multi-resolution location-based training for multi-channel continuous speech separation},
  author = {Hassan Taherian and DeLiang Wang},
  journal= {arXiv preprint arXiv:2301.06458},
  year   = {2023}
}

Comments

Submitted to ICASSP 23

R2 v1 2026-06-28T08:12:39.984Z