RNN-T Models Fail to Generalize to Out-of-Domain Audio: Causes and Solutions
This addresses a critical limitation in automatic speech recognition for real-world applications where training and test data domains differ, though it is incremental as it builds on existing RNN-T models.
The paper tackles the problem of RNN-T models failing to generalize to out-of-domain audio, such as from short to long utterances, and proposes solutions including regularization and dynamic overlapping inference, which improve word error rates from 22.3% to 14.8% and from 67.0% to 25.3% on specific test sets.
In recent years, all-neural end-to-end approaches have obtained state-of-the-art results on several challenging automatic speech recognition (ASR) tasks. However, most existing works focus on building ASR models where train and test data are drawn from the same domain. This results in poor generalization characteristics on mismatched-domains: e.g., end-to-end models trained on short segments perform poorly when evaluated on longer utterances. In this work, we analyze the generalization properties of streaming and non-streaming recurrent neural network transducer (RNN-T) based end-to-end models in order to identify model components that negatively affect generalization performance. We propose two solutions: combining multiple regularization techniques during training, and using dynamic overlapping inference. On a long-form YouTube test set, when the nonstreaming RNN-T model is trained with shorter segments of data, the proposed combination improves word error rate (WER) from 22.3% to 14.8%; when the streaming RNN-T model trained on short Search queries, the proposed techniques improve WER on the YouTube set from 67.0% to 25.3%. Finally, when trained on Librispeech, we find that dynamic overlapping inference improves WER on YouTube from 99.8% to 33.0%.