Leveraging Real Conversational Data for Multi-Channel Continuous Speech Separation
This work addresses the challenge of acquiring high-quality training data for conversational speech separation, which is crucial for improving meeting transcription systems, though it is incremental in nature.
The paper tackles the problem of data mismatch and scarcity in multi-channel continuous speech separation by proposing a three-stage training scheme that leverages both supervised and large-scale unsupervised real conversational data, resulting in steady improvements across training stages on meeting transcription tasks.
Existing multi-channel continuous speech separation (CSS) models are heavily dependent on supervised data - either simulated data which causes data mismatch between the training and real-data testing, or the real transcribed overlapping data, which is difficult to be acquired, hindering further improvements in the conversational/meeting transcription tasks. In this paper, we propose a three-stage training scheme for the CSS model that can leverage both supervised data and extra large-scale unsupervised real-world conversational data. The scheme consists of two conventional training approaches -- pre-training using simulated data and ASR-loss-based training using transcribed data -- and a novel continuous semi-supervised training between the two, in which the CSS model is further trained by using real data based on the teacher-student learning framework. We apply this scheme to an array-geometry-agnostic CSS model, which can use the multi-channel data collected from any microphone array. Large-scale meeting transcription experiments are carried out on both Microsoft internal meeting data and the AMI meeting corpus. The steady improvement by each training stage has been observed, showing the effect of the proposed method that enables leveraging real conversational data for CSS model training.