Deliberation of Streaming RNN-Transducer by Non-autoregressive Decoding
This work addresses speech recognition accuracy for streaming applications, but it is incremental as it builds on existing Align-Refine methods.
The paper tackled the problem of improving streaming RNN-T model accuracy by using non-autoregressive decoding for deliberation, resulting in significantly more accurate recognition results with only a small increase in model parameters.
We propose to deliberate the hypothesis alignment of a streaming RNN-T model with the previously proposed Align-Refine non-autoregressive decoding method and its improved versions. The method performs a few refinement steps, where each step shares a transformer decoder that attends to both text features (extracted from alignments) and audio features, and outputs complete updated alignments. The transformer decoder is trained with the CTC loss which facilitates parallel greedy decoding, and performs full-context attention to capture label dependencies. We improve Align-Refine by introducing cascaded encoder that captures more audio context before refinement, and alignment augmentation which enforces learning label dependency. We show that, conditioned on hypothesis alignments of a streaming RNN-T model, our method obtains significantly more accurate recognition results than the first-pass RNN-T, with only small amount of model parameters.