METHOD FOR VOICE RECOGNITION USING GAUSSIAN POTENTIAL FUNCTION NETWORK ALGORITHM AND LEARNING VECTOR QUANTIZATION ALGORITHM

摘要

PURPOSE: A method for voice recognition using gaussian potential function network algorithm and learning vector quantization algorithm is provided to generate optimal code book for voice recognition. CONSTITUTION: A method for voice recognition using gaussian potential function network algorithm and learning vector quantization algorithm includes a firth through a sixth steps. At the first step(S10), a digital data of characteristic part of digital data is extracted by using SPC(linear prediction coefficient) algorithm. At the second step(S20), an output vector x(t) is generated by using GPFN(Gaussian Potential Function Network) algorithm. At the third step(S30), the output vector x(t) is input to LVQ(learning vector quantization) algorithm and a LVQ algorithm study is performed. At the forth step(S40), an optimal code book is generated. At the fifth step(S50), user's voice of a characteristic feature is recognized. At the sixth step(S60), the voice expressed by the user is recognized.