Pushing the envelope - aside [speech recognition]

By Nelson Morgan, Qifeng Zhu, Andreas Stolcke, K. Sonmez, S. Sivadas, Takahiro Shinozaki, Mari Ostendorf, P. Jain, Hynek Hermansky, D. Ellis and 5 more. 2005

103 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

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

In Signal Processing

Citations in the corpus

Deep Learning: Methods and Applications (2014, background): “In particular, a class of deep neural network models forms the basis of the popular “tandem” approach [262], where the output of the discriminatively learned neural network is treated as part”
A tutorial survey of architectures, algorithms, and applications for deep learning (2014, methodology): “The tandem approach generates discriminative features for an HMM by using the activities from one or more hidden layers of a neural network with various ways of information combination, which can be regarded as a form of discriminative deep architectures [33, 106].”
Deep Neural Networks for Acoustic Modeling in Speech Recognition (2012, methodology): “Instead of replacing the coefficients usually modeled by GMM s, neural networks can also be used to provide additional features for the GMM to model [8], [9], [63].”
Machine Learning Paradigms for Speech Recognition: An Overview (2013, methodology): “Due mainly to the difficulty in learning MLPs, this line of research has been switched to a new direction where the MLP simply produces a subset of " feature vectors " in combination with the traditional features for use in the generative HMM [94].”
Automatic speech emotion recognition using modulation spectral features (2011, background): “The limitations of short-term spectral features for speech recognition, however, are considerable (Morgan et al., 2005).”
Investigation of full-sequence training of deep belief networks for speech recognition (2010, background): “Research in speech recognition has explored layered architectures in the recognizer design for quite some time (e.g., [1][2][3][5][ 15 ][17][18][19][20]), motivated partly by the desire to capitalize on some analogous properties in the human speech generation and perception systems.”
An Overview of Deep-Structured Learning for Information Processing (2011, methodology): “The tandem approach generates features for the HMM for phonetic classification, using one or more hidden layers of neural network with various ways of information combination (Morgan et al., 2005; Morgan 2011).”
Speech Recognition Using Augmented Conditional Random Fields (2009, methodology): “-ACRFs can also take advantage of the TRAP tandem approach [64], [65] as a powerful frontend [ 20 ].”
Current trends in multilingual speech processing (2011, background): “As a result, the use of MLP features has led to exploration of spectro-temporal acoustic features (Morgan et al 2005; Hermansky & Fousek 2005), i.e., features that span across and characterize both time and frequency.”
Context-Dependent Pre-Trained Deep Neural Networks for Large-Vocabulary Speech Recognition (2012, methodology): “Around 1999, the desire to use HMM advances from the speech research community directly without developing replacement techniques and tools contributed to a shift from using neural nets to predict phonetic states to using neural nets to augment features for later use in a conventional GMM-HMM recogn”