A Novel Framework and Training Algorithm for Variable-Parameter Hidden Markov Models

By Dong Yu, Li Deng, Yifan Gong, A. Acero. 2009

43 citations recorded by the source corpus.

How researchers see this paper

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

In Artificial Intelligence

Citations in the corpus

Investigation of full-sequence training of deep belief networks for speech recognition (2010, methodology): “…techniques in speech recognition in the DBN framework, such as the use of object functions better correlating with errors than MMI [7], use of context dependency [1][2], and conditioning the DBN weights on specific auxiliary factors that control the variability in the speech data [21][22].”
Learning in the Deep-Structured Conditional Random Fields (2009, unknown): “…í µí±í µí± í µí±¦ í µí±¡−1 , í µí±¦ í µí±¡ , í µí², í µí±¡ = í µí± í µí± í µí± í µí± í µí±¦ í µí±¡−1 , í µí±¦ í µí±¡ , í µí², í µí±¡ í µí± í µí± í µí±¦ í µí±¡−1 , í µí±¦ í µí±¡ , í µí², í µí±¡ , (5) where í µí± í µí± . is a weight function whose definition can be found in [8][10][11].”
Language recognition using deep-structured conditional random fields (2010, background): “where is a weight function whose definition and calculation method can be found in [9][10][11] and the number needs to be determined based on the amount of training data available.”
Context-Dependent Pre-Trained Deep Neural Networks for Large-Vocabulary Speech Recognition (2012, methodology): “Inspiration for such algorithms may come from the ANN-HMM literature (e.g., [72] and [73]) or the many successful adaptation techniques developed in the past decades for GMM-HMMs (e.g., MLLR [74], MAP [75], joint compensation of distortions [76], variable parameter HMMs [77]).”
An Overview of Noise-Robust Automatic Speech Recognition (2014, methodology): “For example, in [177], piecewise spline interpolation is used to represent the dependency of the HMM parameters on the conditioning parameters.”
KL-divergence regularized deep neural network adaptation for improved large vocabulary speech recognition (2013, methodology): “The potential of CD-DNN-HMMs, however, are yet to be explored.”
Binary coding of speech spectrograms using a deep auto-encoder (2010, background): “Also, it is important to condition the discriminative deep auto-encoder’s parameters on specific variability factors (e.g., [2][13][14]).”
An Overview of Deep-Structured Learning for Information Processing (2011, methodology): “(A variant of this approach has been more recently developed with superior learning techniques for time-varying HMM parameters and with the applications extended to speech recognition robustness; e.g., Yu and Deng, 2009; Yu et al., 2009.”