Guided contrastive self-supervised pre-training for automatic speech recognition
This work addresses the problem of enhancing ASR performance for speech recognition systems, presenting an incremental improvement over existing pre-training methods.
The paper tackled the problem of improving automatic speech recognition (ASR) by proposing Guided Contrastive Predictive Coding (GCPC), a modification of Contrastive Predictive Coding (CPC) that injects prior knowledge during pre-training, resulting in reduced Word Error Rates (WER) by 4.44%, 6.55%, and 15.43% relative on German, French, and English datasets compared to training from scratch.
Contrastive Predictive Coding (CPC) is a representation learning method that maximizes the mutual information between intermediate latent representations and the output of a given model. It can be used to effectively initialize the encoder of an Automatic Speech Recognition (ASR) model. We present a novel modification of CPC called Guided Contrastive Predictive Coding (GCPC). Our proposed method maximizes the mutual information between representations from a prior-knowledge model and the output of the model being pre-trained, allowing prior knowledge injection during pre-training. We validate our method on 3 ASR tasks: German, French and English. Our method outperforms CPC pre-training on all three datasets, reducing the Word Error Rate (WER) by 4.44%, 6.55% and 15.43% relative on the German, French and English (Librispeech) tasks respectively, compared to training from scratch, while CPC pre-training only brings 2.96%, 1.01% and 14.39% relative WER reduction respectively.