Faster, Simpler and More Accurate Hybrid ASR Systems Using Wordpieces
This work addresses the engineering burden and computational cost in ASR systems for speech recognition applications, though it is incremental as it builds on existing transformer and CTC methods.
The paper tackles the complexity and inefficiency of hybrid automatic speech recognition (ASR) systems by using wordpieces with connectionist temporal classification (CTC) training, achieving state-of-the-art results on LibriSpeech and competitive word-error-rates while simplifying the engineering pipeline and improving runtime efficiency.
In this work, we first show that on the widely used LibriSpeech benchmark, our transformer-based context-dependent connectionist temporal classification (CTC) system produces state-of-the-art results. We then show that using wordpieces as modeling units combined with CTC training, we can greatly simplify the engineering pipeline compared to conventional frame-based cross-entropy training by excluding all the GMM bootstrapping, decision tree building and force alignment steps, while still achieving very competitive word-error-rate. Additionally, using wordpieces as modeling units can significantly improve runtime efficiency since we can use larger stride without losing accuracy. We further confirm these findings on two internal VideoASR datasets: German, which is similar to English as a fusional language, and Turkish, which is an agglutinative language.