2011/04/06 by George E. Dahl, Dong Yu, Li Deng +1 · 3 citations
Computer Science · Mathematics · #Speech Recognition and Synthesis #Music and Audio Processing #Speech and Audio Processing #Hidden Markov model #Computer science #Speech recognition #Word error rate #Artificial intelligence #Artificial neural network #Context (archaeology) #Mixture model #Deep neural networks #Sentence #Generalization #Phone #Deep learning #Pattern recognition (psychology) #Vocabulary #Mathematics
paper · doi:10.1109/tasl.2011.2134090
openalex publication_date 2011/04/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/30
We propose a novel context-dependent (CD) model for large-vocabulary speech recognition (LVSR) that leverages recent advances in using deep belief networks for phone recognition. We describe a pre-trained deep neural network hidden Markov model (DNN-HMM) hybrid architecture that trains the DNN to produce a distribution over senones (tied triphone states) as its output. The deep belief network pre-training algorithm is a robust and often helpful way to initialize deep neural networks generatively that can aid in optimization and reduce generalization error. We illustrate the key components of our model, describe the procedure for applying CD-DNN-HMMs to LVSR, and analyze the effects of various modeling choices on performance. Experiments on a challenging business search dataset demonstrate that CD-DNN-HMMs can significantly outperform the conventional context-dependent Gaussian mixture model (GMM)-HMMs, with an absolute sentence accuracy improvement of 5.8% and 9.2% (or relative error reduction of 16.0% and 23.2%) over the CD-GMM-HMMs trained using the minimum phone error rate (MPE) and maximum-likelihood (ML) criteria, respectively.