Stephan Euler

The influence of speech coding algorithms on automatic speech recognition

Examines the influence of different coders in the range from 64 kbit/sec to 4.8 kbit/sec on both a speaker independent isolated word recognizer and a speaker verification system. Applying systems trained with 64 kbit/sec to e.g. the 4.8 kbit/sec data increases the error rate of the word recognizer by a factor of three. For rates below 13 kbit/sec the speaker verification is more affected than the word recognition. The performance improves significantly if word models are provided for the individual coding conditions. Therefore, the authors use a Gaussian classifier for estimation of the coding condition of a test utterance. The combination of this classifier and coder specific word models yields a high overall recognition performance.<<ETX>>

Extending the vocabulary of a speaker independent recognition system

The authors discuss the extension and adaptation of a speaker-independent, small-vocabulary, isolated word recognition system based on tied density hidden Markov models. In the proposed approach, the density functions are trained from a basic set of words using acoustic segmentation, position-dependent segment labeling, and clustering of the segment specific densities. Then the parameters of the word models are estimated by means of a Viterbi update procedure. With a given set of densities the Viterbi update can also be used to generate models for words not included in the basic set. The dependency between the recognition performance and the amount of reference data both for speaker-independent and speaker-dependent experiments is examined in detail. The authors compare different algorithms to avoid zero probabilities in the word models due to insufficient data.<<ETX>>

Comparison of whole word and subword modeling techniques for speaker verification with limited training data

In this paper we use whole word and subword hidden Markov models for text dependent speaker verification. In this application usually only a small amount of training data is available for each model. In order to cope with this limitation we propose a intermediate functional representation of the training data allowing the robust initialization of the models. This new approach is tested with two databases and is compared both with standard training techniques and the dynamic time warp method. Secondly, we give results for two types of subword units. The scores of these units are combined in two different ways to obtain word error rates.

A time continuous model for speech recognition

In this paper we present a time continuous extension of the hidden Markov model approach in order to obtain a better representation of the continuous nature of the speech process. The discrete state sequence of the hidden Markov model is replaced by a continuous parameter, varying between 0 and 1. For an utterance and a given word model an optimum mapping of the feature vectors to the continuous axis is found and the likelihood is calculated based on this mapping. As a first test of this very general approach we extended the hidden Markov model by first mapping the states onto the new axis. Values between the states are then obtained by interpolation between the states. As alternatives we considered interpolation of either the likelihood values of the state density functions and/or of the parameters of the density functions itself. The approach was tested in a speaker independent isolated word recognition system.

Java tools for teaching speech recognition

In this paper, we present our concept for a sequence of experiments with speech recognizers used in teaching speech recognition techniques. The experiments are performed with a combination of own tools and the hidden Markov toolkit (HTK). The first experiment demonstrates speaker dependent recognition based on the dynamic time warp algorithm. In the course of this experiment all utterances from the students are recorded and used to build up a data base. Both the recognizer and the tool used for viewing and editing the speech data are written in Java making them platform independent and easy to extend. The recorded speech data is then utilized to train and test a speaker independent recognizer.