On Modular Training of Neural Acoustics-to-Word Model for LVCSR
This work addresses the problem of reducing data requirements and leveraging traditional techniques in ASR for researchers and practitioners, though it is incremental as it builds on existing modular approaches.
The paper tackles the challenge of training end-to-end automatic speech recognition systems by proposing a modular framework that separately trains neural acoustic and language models, then integrates them, achieving significant improvement over a direct acoustics-to-word model on a 300-hour Switchboard task.
End-to-end (E2E) automatic speech recognition (ASR) systems directly map acoustics to words using a unified model. Previous works mostly focus on E2E training a single model which integrates acoustic and language model into a whole. Although E2E training benefits from sequence modeling and simplified decoding pipelines, large amount of transcribed acoustic data is usually required, and traditional acoustic and language modelling techniques cannot be utilized. In this paper, a novel modular training framework of E2E ASR is proposed to separately train neural acoustic and language models during training stage, while still performing end-to-end inference in decoding stage. Here, an acoustics-to-phoneme model (A2P) and a phoneme-to-word model (P2W) are trained using acoustic data and text data respectively. A phone synchronous decoding (PSD) module is inserted between A2P and P2W to reduce sequence lengths without precision loss. Finally, modules are integrated into an acousticsto-word model (A2W) and jointly optimized using acoustic data to retain the advantage of sequence modeling. Experiments on a 300- hour Switchboard task show significant improvement over the direct A2W model. The efficiency in both training and decoding also benefits from the proposed method.