Titlebook: Speech Recognition and Coding; New Advances and Tre Antonio J. Rubio Ayuso,Juan M. López Soler Conference proceedings 1995 Springer-Verlag

Incompetent

Explicit Modelling of Duration in HMM: an Efficient Algorithmor weakness of HMM is that the state duration probability density functions (SDPDF) are exponential, which is not appropriate for speech signals. In order to cope with this deficiency some authors have proposed to model explicitly the state duration. In these models the first order Markov hypothesis

图画文字

Acoustic-Phonetic Decoding of Spanish Continuous Speech with Hidden Markov Models We will discuss two different choices of sub-lexical units we made for a Spanish decoder phone-like units and discriminative-transitional/steady units. Within the framework of HMM, we will report different series of decoding experiments of Spanish continuous speech. Single and multiple codebook exp

值得

HMM-Based Speech Recognition in Noisy Car Environmentnder the same ambient conditions. The aim of this paper is to report the application of several robust techniques on a system based on the HMM (Hidden Markov Models) and VQ (Vector Quantization) approaches for speech recognition in noisy car environment.

串通

失眠症

刚开始

On the Performance of SCHMM for Isolated Word Recognition and Rejectiond will be recognised as one of the keywords, thus steering the dialogue into a wrong direction. The use of garbage or sink models (SM) is a known technique to avoid those extraneous words being recognised as vocabulary words. Each lexical word from the active vocabulary is represented in the recogni

CRUDE

The Speech Recognition Research System of the TU Dresdenology for research purposes. The main topics of interest with this system are the use of knowledge on human speech perception, the event-directed recognition strategy, and the use of syllables as basic units of speech production and perception. The system is implemented on DECstations 5000/125 in th

折磨

Leisureliness

Evaluation of ASR Systems, Algorithms and Databasess. Techniques to model the variability of speech are proposed. They are implemented on a test workstation. Predicting the performance of a given recognizer in a particular situation is possible. It is argued that most of these techniques could also be adapted to improve the robustness of recognizers and speaker verifiers.

完整

Heuristic Search Methods for a Segment Based Continuous Speech Recognizerased Neural Network / Dynamic Programming hybrid [1], we have therefore introduced a heuristic search method. It is shown that this search is significantly faster than a Viterbi beam search, while there is no degradation of the recognition results.