AI summary

This paper introduces a new ASR system that uses DBN-pretrained ANN/HMM models, trained on large datasets of Voice Search and YouTube data, to outperform GMM/HMM baselines by 3.7% and 4.7% absolute WER, respectively, with further gains from MMI fine-tuning and SCARF model combination.

Abstract

The use of Deep Belief Networks (DBN) to pretrain Neural Networks has recently led to a resurgence in the use of Artificial Neural Network Hidden Markov Model (ANN/HMM) hybrid systems for Automatic Speech Recognition (ASR). In this paper we report results of a DBN-pretrained context-dependent ANN/HMM system trained on two datasets that are much larger than any reported previously with DBN-pretrained ANN/HMM systems - 5870 hours of Voice Search and 1400 hours of YouTube data. On the first dataset, the pretrained ANN/HMM system outperforms the best Gaussian Mixture Model - Hidden Markov Model (GMM/HMM) baseline, built with a much larger dataset by 3.7% absolute WER, while on the second dataset, it outperforms the GMM/HMM baseline by 4.7% absolute. Maximum Mutual Information (MMI) fine tuning and model combination using Segmental Conditional Random Fields (SCARF) give additional gains of 0.1% and 0.4% on the first dataset and 0.5% and 0.9% absolute on the second dataset.