Phoneme Based Neural Transducer for Large Vocabulary Speech Recognition
This work addresses speech recognition for large vocabularies, presenting an incremental improvement by integrating phoneme-level modeling with neural networks.
The authors tackled the problem of large vocabulary speech recognition by proposing a phoneme-based neural transducer that combines classical and end-to-end approaches, achieving performance comparable to state-of-the-art results on the TED-LIUM Release 2 and Switchboard corpora.
To join the advantages of classical and end-to-end approaches for speech recognition, we present a simple, novel and competitive approach for phoneme-based neural transducer modeling. Different alignment label topologies are compared and word-end-based phoneme label augmentation is proposed to improve performance. Utilizing the local dependency of phonemes, we adopt a simplified neural network structure and a straightforward integration with the external word-level language model to preserve the consistency of seq-to-seq modeling. We also present a simple, stable and efficient training procedure using frame-wise cross-entropy loss. A phonetic context size of one is shown to be sufficient for the best performance. A simplified scheduled sampling approach is applied for further improvement and different decoding approaches are briefly compared. The overall performance of our best model is comparable to state-of-the-art (SOTA) results for the TED-LIUM Release 2 and Switchboard corpora.