PATTERN RECOGNITION SYSTEM

摘要

A pattern recognition method of dynamic time warping of two sequences of feature sets onto each other for use in a distance pattern recognition system is provided. The system converts the audio pattern (50) to be analyzed into a feature set in an integer format using processor (52). The integer feature set for the test pattern is then compared with a database (54) containing reference patterns, by measuring the spectral distance between the test pattern and each reference pattern, using dynamic time warping unit (56). The processor (52) digitizes audio pattern (50) using an analog-to-digital converter (60) and then detects the endpoints of each audio pattern (50) using detector (62). The output is an audio word. The processor breaks the word into frames and then extracts the features of each frame via a feature extractor (64). Feature extractor (64) comprises a linear prediction coefficient calculator (66), a cepstrum converter (68) and an integer formatter (70).