Improving RNN-Transducers with Acoustic LookAhead
This addresses hallucination issues in speech-to-text models, improving accuracy for streaming applications, but it is incremental as it builds on existing RNN-T architectures.
The paper tackles the problem of multi-step hallucination in RNN-Transducers due to LM biasing by proposing LookAhead, which grounds text representations in future audio input, resulting in a 5%-20% relative reduction in word error rate.
RNN-Transducers (RNN-Ts) have gained widespread acceptance as an end-to-end model for speech to text conversion because of their high accuracy and streaming capabilities. A typical RNN-T independently encodes the input audio and the text context, and combines the two encodings by a thin joint network. While this architecture provides SOTA streaming accuracy, it also makes the model vulnerable to strong LM biasing which manifests as multi-step hallucination of text without acoustic evidence. In this paper we propose LookAhead that makes text representations more acoustically grounded by looking ahead into the future within the audio input. This technique yields a significant 5%-20% relative reduction in word error rate on both in-domain and out-of-domain evaluation sets.