Approximate Nearest Neighbour Phrase Mining for Contextual Speech Recognition
This work addresses the challenge of improving biasing accuracy in speech recognition systems when faced with similar phrases, offering an incremental enhancement to existing methods.
The paper tackles the problem of disambiguating similar phrases in contextual speech recognition by mining hard negative phrases using approximate nearest neighbor search during training, resulting in up to 7% relative word error rate reductions for the contextual portion of test data.
This paper presents an extension to train end-to-end Context-Aware Transformer Transducer ( CATT ) models by using a simple, yet efficient method of mining hard negative phrases from the latent space of the context encoder. During training, given a reference query, we mine a number of similar phrases using approximate nearest neighbour search. These sampled phrases are then used as negative examples in the context list alongside random and ground truth contextual information. By including approximate nearest neighbour phrases (ANN-P) in the context list, we encourage the learned representation to disambiguate between similar, but not identical, biasing phrases. This improves biasing accuracy when there are several similar phrases in the biasing inventory. We carry out experiments in a large-scale data regime obtaining up to 7% relative word error rate reductions for the contextual portion of test data. We also extend and evaluate CATT approach in streaming applications.