Serialized Output Training for End-to-End Overlapped Speech Recognition
This addresses the problem of transcribing multi-speaker overlapped audio for speech recognition systems, offering a more flexible and efficient approach compared to existing methods.
The paper tackles overlapped speech recognition by proposing serialized output training (SOT), a framework that generates transcriptions for multiple speakers sequentially using a single output layer, and it outperforms permutation invariant training (PIT)-based models on the LibriSpeech corpus with variable speaker counts.
This paper proposes serialized output training (SOT), a novel framework for multi-speaker overlapped speech recognition based on an attention-based encoder-decoder approach. Instead of having multiple output layers as with the permutation invariant training (PIT), SOT uses a model with only one output layer that generates the transcriptions of multiple speakers one after another. The attention and decoder modules take care of producing multiple transcriptions from overlapped speech. SOT has two advantages over PIT: (1) no limitation in the maximum number of speakers, and (2) an ability to model the dependencies among outputs for different speakers. We also propose a simple trick that allows SOT to be executed in $O(S)$, where $S$ is the number of the speakers in the training sample, by using the start times of the constituent source utterances. Experimental results on LibriSpeech corpus show that the SOT models can transcribe overlapped speech with variable numbers of speakers significantly better than PIT-based models. We also show that the SOT models can accurately count the number of speakers in the input audio.