A Comparative Study of Modular and Joint Approaches for Speaker-Attributed ASR on Monaural Long-Form Audio
This addresses the challenge of accurately recognizing 'who spoke what' in multi-talker recordings for applications like meeting transcription, though it is incremental as it builds on existing modular and joint methods.
The paper tackled the problem of speaker-attributed automatic speech recognition (SA-ASR) on real monaural long-form audio by comparing modular and joint approaches, finding that after fine-tuning with small real data, the joint system performed 8.9–29.9% better in accuracy than the best modular system.
Speaker-attributed automatic speech recognition (SA-ASR) is a task to recognize "who spoke what" from multi-talker recordings. An SA-ASR system usually consists of multiple modules such as speech separation, speaker diarization and ASR. On the other hand, considering the joint optimization, an end-to-end (E2E) SA-ASR model has recently been proposed with promising results on simulation data. In this paper, we present our recent study on the comparison of such modular and joint approaches towards SA-ASR on real monaural recordings. We develop state-of-the-art SA-ASR systems for both modular and joint approaches by leveraging large-scale training data, including 75 thousand hours of ASR training data and the VoxCeleb corpus for speaker representation learning. We also propose a new pipeline that performs the E2E SA-ASR model after speaker clustering. Our evaluation on the AMI meeting corpus reveals that after fine-tuning with a small real data, the joint system performs 8.9--29.9% better in accuracy compared to the best modular system while the modular system performs better before such fine-tuning. We also conduct various error analyses to show the remaining issues for the monaural SA-ASR.