Основні підходи до розпізнавання мовленнєвої інформації (частина 1)

Abstract

This paper is devoted to describing and classification of the nowadays available speech processing and recognition techniques. Initial stages of speech signal processing are investigated and speech recognition systems are classified. Various approaches of processing, treatment and recognition of speech signal are analyzed. Speech signal noise reduction, segmentation and parameterization methods are also come into question. Great attention dedicated to the comparison of the different methods of speech recognition. Correlation analysis, phoneme-reference method, expert systems, hidden Markov models and Gaussian mixture models are considered.