Hybrid HMM–NN modeling of stationary–transitional units for continuous speech recognition

Albesano, D.; Gemello, Roberto; Mana, Franco

doi:10.1016/s0020-0255(99)00106-1

Cited by 24 publications

(15 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the experiments reported in this paper, a single phonetic decoder has been used, i.e. the Loquendo-ASR recognizer for Italian language [15], which generates a lattice of hypotheses. We estimate n-gram statistics up to the third order from these hypotheses leading to a 44135-dimensional feature space.…”

Section: A Phonetic Modelsmentioning

confidence: 99%

Analysis of Large-Scale SVM Training Algorithms for Language and Speaker Recognition

Cumani

Laface

2012

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

This paper compares a set of large scale Support Vector Machine training algorithms for language and speaker recognition tasks. We analyze five approaches for training phonetic and acoustic SVM models for language recognition. We compare the performance of these approaches as a function of the training time required by each of them to reach convergence, and we discuss their scalability towards large corpora. Two of these algorithms can be used in speaker recognition to train a SVM that classifies pairs of utterances as either belonging to the same speaker or to two different speakers. Our results show that the accuracy of these algorithms is asymptotically equivalent, but they have different behavior with respect to the time required to converge. Some of these algorithms not only scale linearly with the training set size, but are also able to give their best results after just a few iterations. State-of-the-art performance has been obtained in the female subset of the NIST 2010 Speaker Recognition Evaluation extended core test using a single SVM system.

show abstract

Section: A Phonetic Modelsmentioning

confidence: 99%

Analysis of Large-Scale SVM Training Algorithms for Language and Speaker Recognition

Cumani

Laface

2012

IEEE Trans. Audio Speech Lang. Process.

View full text Add to dashboard Cite

show abstract

“…NNs are commonly applied to voice and language related problems, for example McNamara et al (1998) and Albesano et al (2000). Speech recognition requires the ability to deal with temporal sequences, for which a variety of methods have been applied.…”

Section: Neural Networkmentioning

confidence: 99%

The state of play in machine/environment interactions

Aitkenhead

McDonald

2006

Artif Intell Rev

View full text Add to dashboard Cite

Due to the breadth of the subject, it is no longer possible to provide a review of all of the work being carried out in the field of Artificial Intelligence. However, a more localised review of research taking place in the overlap between engineering, AI and psychology can be meaningfully performed. We show here that while there have been marked successes in the past few years, there is an identifiable set of 'classic' problems that remain to be solved, and which largely direct the work ongoing in this area. This review aims to discuss the directions being taken at the current time, in particular the developing and maturing possibilities provided by neural networks and evolutionary computation, and by the use of our knowledge of the mind in developing artificial agents capable of mimicking our abilities to interact with the environment.

show abstract

“…The Loquendo-ASR system uses acoustic models based on a hybrid combination of Hidden Markov Models (HMM) and Multi Layer Perceptron (MLP), where each phonetic unit is described in terms of a single or double state left-to-right automaton with self-loops, the HMM transition probabilities are uniform and fixed, and the emission probabilities are computed by an MLP [16]. This MLP has an input layer of 273 units, a first hidden layer of 315 units, a second hidden layer of 315 units and an output layer including a variable number of units that is language dependent (600 to 1000).…”

Section: A the Ann Architecturementioning

confidence: 99%

“…The acoustic models are based on a set of vocabulary and gender independent units including stationary contextindependent phones and diphone-transition coarticulation models [16]. These acoustic models have been successfully used for the 15 languages released with the Loquendo ASR recognizer, and are the seed models for the adaptation experiments of Section V, if not differently specified.…”

Section: A the Ann Architecturementioning

confidence: 99%

Adaptation of Artificial Neural Networks Avoiding Catastrophic Forgetting

Albesano¹,

Gemello²,

Laface

et al. 2006

The 2006 IEEE International Joint Conference on Neural Network Proceedings

View full text Add to dashboard Cite

In connectionist learning, one relevant problem is "catastrophic forgetting" that may occur when a network, trained with a large set of patterns, has to learn new input patterns, or has to be adapted to a different environment. The risk of catastrophic forgetting is particularly high when a network is adapted with new data that do not adequately represent the knowledge included in the original training data.Two original solutions are proposed to reduce the risk that the network focuses on new data only, loosing its generalization capability. The first one, Conservative Training, is a variant to the target assignment policy, while the second approach, Support Vector Rehearsal, selects from the training set the patterns that lay near the borders of the classes not included in the adaptation set. These patterns are used as sentinels that try to keep unchanged the original boundaries of these classes. Moreover, we investigated the extension of the classical approach consisting in applying linear transformations not only to the input features, but also to the outputs of the internal layers. The motivation is that the outputs of an internal layer represent a projection of the input pattern into a space where it should be easier to learn the classification or transformation expected at the output of the network.We illustrate the problems using an artificial test-bed, and apply our techniques to a set of adaptation tasks in the domain of Automatic Speech Recognition (ASR) based on Artificial Neural Networks. Supervised ASR adaptation experiments with several corpora and for different adaptation types are described. We report on the adaptation potential of different techniques, and on the generalization capability of the adapted networks. The results show that the combination of the proposed approaches mitigates the catastrophic forgetting effects, and always outperforms the use of the classical transformations in the feature space.

show abstract

Hybrid HMM–NN modeling of stationary–transitional units for continuous speech recognition

Cited by 24 publications

References 8 publications

Analysis of Large-Scale SVM Training Algorithms for Language and Speaker Recognition

Analysis of Large-Scale SVM Training Algorithms for Language and Speaker Recognition

The state of play in machine/environment interactions

Adaptation of Artificial Neural Networks Avoiding Catastrophic Forgetting

Contact Info

Product

Resources

About