Joint Speaker Counting, Speech Recognition, and Speaker Identification for Overlapped Speech of Any Number of Speakers
arXiv:2006.10930
Abstract
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.
Accepted to INTERSPEECH 2020
References in corpus (7)
- SpecAugment: A Simple Data Augmentation Method for Automatic Speech Recognition
- VoxCeleb2: Deep Speaker Recognition
- Attention-Based Models for Speech Recognition
- Deep attractor network for single-microphone speaker separation
- End-to-end Continuous Speech Recognition using Attention-based Recurrent NN: First Results
- Semantic Mask for Transformer based End-to-End Speech Recognition
- Advances in Online Audio-Visual Meeting Transcription