Improvements to Embedding-Matching Acoustic-to-Word ASR Using Multiple-Hypothesis Pronunciation-Based Embeddings.

Hao Yen, Woojay Jeon

Browse the full ICASSP paper archive.