Active and Semi-Supervised Learning in ASR: Benefits on the Acoustic and Language Models
This is an incremental improvement for speech recognition systems, focusing on cost reduction and performance gains in acoustic and language models.
The paper tackles the problem of reducing transcription costs in automatic speech recognition by jointly applying active learning and semi-supervised training, showing that this approach reduces transcription costs by about 70% over random selection or improves word error rate by about 12.5% relative for a fixed budget.
The goal of this paper is to simulate the benefits of jointly applying active learning (AL) and semi-supervised training (SST) in a new speech recognition application. Our data selection approach relies on confidence filtering, and its impact on both the acoustic and language models (AM and LM) is studied. While AL is known to be beneficial to AM training, we show that it also carries out substantial improvements to the LM when combined with SST. Sophisticated confidence models, on the other hand, did not prove to yield any data selection gain. Our results indicate that, while SST is crucial at the beginning of the labeling process, its gains degrade rapidly as AL is set in place. The final simulation reports that AL allows a transcription cost reduction of about 70% over random selection. Alternatively, for a fixed transcription budget, the proposed approach improves the word error rate by about 12.5% relative.