National language single tone recognizing system capable of wide application

Abstract

The invention relates to a widely applicable mother tongue monosyllable identification system which comprises a previous processor, a characteristic matrix method in which the known monosyllable sound wave normalization is consistent with the selected size, wherein, an elastic frame is adopted to normalize the sound wave and transfer the sound wave into linear prediction coding cepstrum characteristic matrixes in equal size and the same known monosyllable sound wave is transferred into matrixes with the same characteristics; to each known monosyllable, K best samples are chosen; and the K best samples of one known monosyllable characteristic matrix are transferred into a standard model and stored in a database. The standard model comprises the mean value and the variance number of the K best samples of the known monosyllable characteristic matrix; the unknown monosyllable sound wave is normalized and transferred into a characteristic matrix which has the same size with the known standard model, includes the linear prediction coding ceptrum inside, and is called as an unknown monosyllable classifying model; the unknown monosyllable classifying model are compared with all the known monosyllable standard model in the database to find a known monosyllable, which has the smallest distance with the unknown monosyllable and is identified as an unknown monosyllable.

Description

technical field [0001] What the invention relates to is a kind of voice time difference system. Background technique [0002] When a single sound is pronounced, its pronunciation is represented by a sound wave. A sound wave is a system that changes nonlinearly with time. A single sound wave contains a dynamic characteristic that also changes nonlinearly and continuously with time. When the same single sound is pronounced, there is a series of the same dynamic characteristics, which expand and contract nonlinearly with time, but the same dynamic characteristics are arranged in the same order according to time, but the time is different. When the same single sound is pronounced, it is very difficult to arrange the same dynamic characteristics at the same time position. Also because there are so many similar monophonic sounds, it is difficult to identify them. [0003] A computerized language recognition system first needs to extract the language information related to the s...

AI Technical Summary

Problems solved by technology

If there are differences in time changes of the same pronunciation, one side compares and one side pulls the same feature to the same time position, the recognition rate will be good, but it is difficult to pull the same feature to the same position and the distortion time is too long to be applied

The vector quantization method is not only inaccurate but also time-consuming to identify a large number of single tones

The Hidden Markov Model (HMM) identification method is good, but the method is complicated, too many unknown parameters need to be estimated, and it takes time to calculate the estimated value and identify

Recently, the Bayesian classification method [2] was used to compress various long and short linear predictive coded cepstrum (LPCC) vectors into a classification model of the same size with the same database, and the identification results were better than those of the hidden Markov model method (HMM). The method is good [2, 5], but the compression process is complicated and time-consuming, and it is difficult to compress the same features to the same time position for the same tone, and it is difficult to identify similar tones

Images