Title :
Speech signal processing in order to increase recognition of spoken language
Author :
Halas, Heiiryk R.
Author_Institution :
Inst. Automatyki i Robotyki, Politech. Warszawskiej, Poland
Abstract :
Describes a method of pre-processing speech signal in order to increase phoneme recognition accuracy. The recognition of phonemes is performed by a neural network. Recognition is based on features of speech signal. The features have been selected in order to increase separability of phoneme classes and the recognition level. The proposed method can be easily applied to continuous multispeaker speech recognition system built of neural network and discrete hidden Markov models
Keywords :
hidden Markov models; neural nets; speaker recognition; speech processing; continuous multispeaker speech recognition system; discrete hidden Markov models; neural network; phoneme recognition accuracy; pre-processing method; recognition level; separability; speech signal processing; spoken language recognition; Automatic speech recognition; Frequency; Loudspeakers; Natural languages; Neural networks; Robotics and automation; Signal processing; Speech processing; Speech recognition; Vectors;
Conference_Titel :
Electronics, Circuits and Systems, 1999. Proceedings of ICECS '99. The 6th IEEE International Conference on
Conference_Location :
Pafos
Print_ISBN :
0-7803-5682-9
DOI :
10.1109/ICECS.1999.812279
