Large Scale Language Modeling in Automatic Speech Recognition
This work addresses speech recognition accuracy for applications like search and transcription, but it is incremental as it builds on existing methods with larger data and models.
The paper tackled improving automatic speech recognition by scaling up language models and training data, achieving relative word error rate reductions of 6% to 10% across tasks like Voice Search and YouTube transcription.
Large language models have been proven quite beneficial for a variety of automatic speech recognition tasks in Google. We summarize results on Voice Search and a few YouTube speech transcription tasks to highlight the impact that one can expect from increasing both the amount of training data, and the size of the language model estimated from such data. Depending on the task, availability and amount of training data used, language model size and amount of work and care put into integrating them in the lattice rescoring step we observe reductions in word error rate between 6% and 10% relative, for systems on a wide range of operating points between 17% and 52% word error rate.