Improving RNN Transducer Based ASR with Auxiliary Tasks
This work addresses speech recognition accuracy for applications like social media videos, but it is incremental as it builds on existing RNN-T methods.
The paper tackled improving RNN-T based ASR by using auxiliary tasks, such as context-dependent graphemic state prediction, and achieved competitive results with 2.0%/4.2% WER on LibriSpeech test-clean/other.
End-to-end automatic speech recognition (ASR) models with a single neural network have recently demonstrated state-of-the-art results compared to conventional hybrid speech recognizers. Specifically, recurrent neural network transducer (RNN-T) has shown competitive ASR performance on various benchmarks. In this work, we examine ways in which RNN-T can achieve better ASR accuracy via performing auxiliary tasks. We propose (i) using the same auxiliary task as primary RNN-T ASR task, and (ii) performing context-dependent graphemic state prediction as in conventional hybrid modeling. In transcribing social media videos with varying training data size, we first evaluate the streaming ASR performance on three languages: Romanian, Turkish and German. We find that both proposed methods provide consistent improvements. Next, we observe that both auxiliary tasks demonstrate efficacy in learning deep transformer encoders for RNN-T criterion, thus achieving competitive results - 2.0%/4.2% WER on LibriSpeech test-clean/other - as compared to prior top performing models.