An investigation of phone-based subword units for end-to-end speech recognition
Preprint
Abstract
Phones and their context-dependent variants have been the standard modeling units for conventional speech recognition systems, while characters and character-based subwords are becoming increasingly popular for end-to-end recognition systems. We investigate the use of phone-based subwords, and byte pair encoding (BPE) in particular, as modeling units for end-to-end speech recognition, and develop multi-level language model-based decoding algorithms based on a pronunciation dictionary. Besides the use of the lexicon which is easily available, our system avoids the need of additional expert knowledge or processing steps from conventional systems. Experimental results show that phone-based BPEs lead to more accurate recognition systems than the character-based counterpart, and further improvement can be obtained with the newly developed one-pass beam search decoder, which efficiently combines both phone-based and character-based BPE systems. For Switchboard, our phone-based BPE system achieves 6.8%/14.4% word error rate (WER) on the Switchboard/CallHome portion of the test set while the ensemble system achieves 6.3%/13.3% WER.Keywords
All Related Versions
This publication has 28 references indexed in Scilit:
- Espresso: A Fast End-to-End Neural Speech Recognition Toolkit2019
- A Comparative Study on Transformer vs RNN in Speech Applications2019
- Multi-level language modeling and decoding for open vocabulary end-to-end speech recognition2017
- Exploring architectures, data and units for streaming end-to-end speech recognition with RNN-transducer2017
- An empirical exploration of CTC acoustic models2016
- EESEN: End-to-end speech recognition using deep RNN models and WFST-based decoding2015
- Grapheme-to-phoneme conversion using Long Short-Term Memory recurrent neural networks2015
- Japanese and Korean voice search2012
- Joint-sequence models for grapheme-to-phoneme conversionSpeech Communication, 2008
- SWITCHBOARD: telephone speech corpus for research and development1992