Phonemic hidden Markov models with continuous mixture output densities for large vocabulary word recognition

By Li Deng, Patrick Kenny, Matthew Lennig, Vishwa Gupta, F. Seitz, P. Mermelstein. 1991

The authors demonstrate the effectiveness of phonemic hidden Markov models with Gaussian mixture output densities (mixture HMMs) for speaker-dependent large-vocabulary word recognition. Speech recognition experiments show that for almost any reasonable amount of training data, recognizers using mixture HMMs consistently outperform those employing unimodal Gaussian HMMs. With a sufficiently large training set (e.g. more than 2500 words), use of HMMs with 25-component mixture distributions typically reduces recognition errors by about 40%. It is also found that the mixture HMMs outperform a set of unimodal generalized triphone models having the same number of parameters. Previous attempts to employ mixture HMMs for speech recognition proved discouraging because of the high complexity and computational cost in implementing the Baum-Welch training algorithm. It is shown how mixture HMMs can be implemented very simply in unimodal transition-based frameworks by allowing multiple transitions from one state to another. >

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

1. Jiuxiang Gu
2. Zidong Wang
3. Jianfei Cai
4. Yurong Liu
5. Xiaohui Liu
6. Amir Shahroudy
7. Jun Liu
8. Jason Kuen

Citations in the corpus

• Recent advances in convolutional neural networks (2015, methodology): “is the technology that converts human speech into spoken words. Before applying CNN to ASR, this domain has long been dominated by the Hidden Markov Model and Gaussian Mixture Model (GMM-HMM) methods [298, 299] , which usually require extracting hand-craft features on speech signals, e.g ., the most”
• A survey of deep neural network architectures and their applications (2017, background): “Among them, HMMs and GMMs are widely used for acoustic modeling to generate low-level acoustic contents from high-level acoustic inputs [32,34,79,101,116,120,130].”
• Use of generalized dynamic feature parameters for speech recognition: maximum likelihood and minimum classification error approaches (1995, background): “DURING THE past decade, use of the dynamic feature parameters associated with speech spectra has resulted in demonstrable success in enhancing the performance of speech recognition systems [2], [9], [11], [12], [15], [16], [17].”
• HMM-based speech recognition using state-dependent, discriminatively derived transforms on mel-warped DFT features (1997, unknown): “…on the MFB log channel energies) by taking the difference between two frame forward and two frame backward of the related static features according to [8] (2) This fixed window length of four (equivalent to 40 ms) appears to have reasonably well captured the slope of the spectral envelope.”