Fast voice identifying method for Chinese phrase of specific person

Abstract

The present invented method comprises speech sampling and preprocessing, speech acoustic feature extraction, syllable fuzzy division, speech template training, speech pattern matching and recognitiondecision. Said method adopts uniquie quick method for processing signal to support Chinese phrase recognition, and processes the advantages of less resource consumption, quick operation speed and high recognition rate. Said method is applicable to the products of toy, personal digital processing device and communication terminal, etc.

Description

technical field [0001] The invention belongs to the technical field of speech recognition, in particular to a fast method for speech recognition of Chinese phrases of a specific person. Background technique [0002] The existing specific person speech recognition method and application system usually have the following deficiencies: [0003] (1) Too many training times or too little recognizable vocabulary, making it inconvenient for users to use; [0004] (2) It supports the recognition of short words (usually less than 4 syllables), but does not support the recognition of phrases (usually about 10 syllables), and its application fields are limited; [0005] (3) The core algorithm adopts the Hidden Markov Model (HMM) or Dynamic Time Warping (DTW) method, which has high complexity, takes up a lot of system resources, and has a slow recognition speed. [0006] Such as Philips, Samsung, and Siemens released in the first half of 2001 have voice dialing function mobile phones ...

AI Technical Summary

Problems solved by technology

[0003] (1) Too many training times or too little recognizable vocabulary, making it inconvenient for users to use;

[0004] (2) It supports the recognition of short words (usually less than 4 syllables), but does not support the recognition of phrases (usually about 10 syllables), and its application fields are limited;

[0005] (3) The core algorithm adopts the Hidden Markov Model (HMM) or Dynamic Time Warping (DTW) method, which has high complexity, takes up a lot of system resources, and has a slow recognition speed

Another example is Tsinghua University's patent "small vocabulary speech recognition method and module thereof" (patent publication number: CN1262502A) and "specific person's speech recognition and voice playback method based on a special chip for speech recognition" (patent publication number: CN1268732A). The traditional speech recognition method has made some improvements, but the specific person speech recognition method still requires the user to train twice or more, and because the HMM is used as the core algorithm, the amount of training and calculation is still very large

Images