End-to-end Multichannel Speaker-Attributed ASR: Speaker Guided Decoder and Input Feature Analysis
Abstract
We present an end-to-end multichannel speaker-attributed automatic speech recognition (MC-SA-ASR) system that combines a Conformer-based encoder with multi-frame crosschannel attention and a speaker-attributed Transformer-based decoder. To the best of our knowledge, this is the first model that efficiently integrates ASR and speaker identification modules in a multichannel setting. On simulated mixtures of LibriSpeech data, our system reduces the word error rate (WER) by up to 12% and 16% relative compared to previously proposed single-channel and multichannel approaches, respectively. Furthermore, we investigate the impact of different input features, including multichannel magnitude and phase information, on the ASR performance. Finally, our experiments on the AMI corpus confirm the effectiveness of our system for real-world multichannel meeting transcription.
Cite
@article{arxiv.2310.10106,
title = {End-to-end Multichannel Speaker-Attributed ASR: Speaker Guided Decoder and Input Feature Analysis},
author = {Can Cui and Imran Ahamad Sheikh and Mostafa Sadeghi and Emmanuel Vincent},
journal= {arXiv preprint arXiv:2310.10106},
year = {2023}
}
Comments
2023 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU 2023), Dec 2023, Taipei, Taiwan