Improving End-to-end Speech Recognition with Pronunciation-assisted Sub-word Modeling
Preprint
Abstract
Most end-to-end speech recognition systems model text directly as a sequence of characters or sub-words. Current approaches to sub-word extraction only consider character sequence frequencies, which at times produce inferior sub-word segmentation that might lead to erroneous speech recognition output. We propose pronunciation-assisted sub-word modeling (PASM), a sub-word extraction method that leverages the pronunciation information of a word. Experiments show that the proposed method can greatly improve upon the character-based baseline, and also outperform commonly used byte-pair encoding methods.Keywords
All Related Versions
This publication has 4 references indexed in Scilit:
- Improved Training of End-to-end Attention Models for Speech Recognition2018
- ESPnet: End-to-End Speech Processing Toolkit2018
- On Modular Training of Neural Acoustics-to-Word Model for LVCSR2018
- Hybrid CTC/Attention Architecture for End-to-End Speech RecognitionIEEE Journal of Selected Topics in Signal Processing, 2017