Joint Speaker Counting, Speech Recognition, and Speaker Identification for Overlapped Speech of Any Number of Speakers
This addresses the challenge of processing overlapped speech in applications like meeting transcription, though it is incremental as it builds on existing serialized output training methods.
The paper tackled the problem of recognizing and identifying speakers in overlapped speech with an arbitrary number of speakers, proposing an end-to-end model that unifies speaker counting, speech recognition, and speaker identification, achieving significantly better speaker-attributed word error rate than baseline methods.
We propose an end-to-end speaker-attributed automatic speech recognition model that unifies speaker counting, speech recognition, and speaker identification on monaural overlapped speech. Our model is built on serialized output training (SOT) with attention-based encoder-decoder, a recently proposed method for recognizing overlapped speech comprising an arbitrary number of speakers. We extend SOT by introducing a speaker inventory as an auxiliary input to produce speaker labels as well as multi-speaker transcriptions. All model parameters are optimized by speaker-attributed maximum mutual information criterion, which represents a joint probability for overlapped speech recognition and speaker identification. Experiments on LibriSpeech corpus show that our proposed method achieves significantly better speaker-attributed word error rate than the baseline that separately performs overlapped speech recognition and speaker identification.