Hidden Markov model speech recognition systems typically use Gaussian mixture models to estimate the distributions of decorrelated acoustic feature vectors that correspond to individual subword units. By contrast, hybrid connectionist-HMM systems use discriminatively-trained neural networks

ABSTRACT Hidden Markov model speech recognition systems typically use Gaussian mixture models to estimate the distributions of decorrelated acoustic feature vectors that correspond to individual subword units. By contrast, hybrid connectionist-HMM systems use discriminatively-trained neural

, however many applications require the features in tandem. Fundamentally, lineage enables simple and consistent representation of uncertain data, it correlates uncertainty in query results with uncertainty in the input data, and query processing with lineage and uncertainty together presents computational

We present a hierarchical architecture for tandem acoustic modeling. In the tandem acoustic modeling paradigm a Multi Layer Perceptron (MLP) is discriminatively trained to estimate phoneme posterior probabilities on a labeled database. The outputs of the MLP after nonlinear transformation

Hidden Markov model speech recognition systems typically use Gaussian mixture models to estimate the distributions of decorrelated acoustic feature vectors that correspond to individual subword units. By contrast, hybrid connectionist-HMM systems use discriminatively-trained neural networks

of performance, by incorporating feature extraction using the fast Fourier transform. The results are compared with that in earlier work to demonstrate its effectiveness. Index Terms—Acoustic remote sensing, classification, fast Fourier transform (FFT), neural networks, sonic detection and

mode, in a supervised or unsupervised mode, and facilitate data and knowledge integration, rule extraction and rule manipulation. The evolving processes of ECOS are defined by parameters, “genes ” [1]. Both ECOS parameters and the input variables (features) for a particular problem are optimized

to posterior probabilities has enabled us to develop a number of novel approaches to confidence estimation, pronunciation modelling and search. In addition we have investigated a new feature extraction technique based on the modulation-filtered spectrogram (MSG), and methods for combining multiple information

A novel connectionist method is proposed to simultaneously use diverse features in an optimal way for pattern classification. Unlike methods of combining multiple classifiers, a modular neural network architecture is proposed through use of soft competition among diverse features. Parameter

Abstract not found