Edit Distance based RL for RNNT decoding
This addresses a key limitation in RNN-T models for ASR, improving accuracy in transcription tasks, though it is an incremental improvement over existing methods.
The paper tackles the discrepancy between training and inference objectives in RNN-T models for automatic speech recognition, proposing an Edit Distance based Reinforcement Learning method that achieved state-of-the-art word error rates on LibriSpeech with a 600M Conformer RNN-T model.
RNN-T is currently considered the industry standard in ASR due to its exceptional WERs in various benchmark tests and its ability to support seamless streaming and longform transcription. However, its biggest drawback lies in the significant discrepancy between its training and inference objectives. During training, RNN-T maximizes all alignment probabilities by teacher forcing, while during inference, it uses beam search which may not necessarily find the maximum probable alignment. Additionally, RNN-T's inability to experience mistakes during teacher forcing training makes it more problematic when a mistake occurs in inference. To address this issue, this paper proposes a Reinforcement Learning method that minimizes the gap between training and inference time. Our Edit Distance based RL (EDRL) approach computes rewards based on the edit distance, and trains the network at every action level. The proposed approach yielded SoTA WERs on LibriSpeech for the 600M Conformer RNN-T model.