VOICE AND TONE CODING METHOD

摘要

PURPOSE: A voice and tone coding method is provided to reduce a distortion at a dual tone multi frequency(DTMF) transfer remarkably by modifying a structure of a coder. CONSTITUTION: In a voice and tone coding method, in order to process an input signal by a sub frame unit, an input signal passing through a high band pass filter is multiplied with a hamming window having 180 samples(601,602). Own correlation function(Rx) is obtained with regard to each sub frame(603), and a binominal window function is multiplied with the function(Rx). The correlation function not multiplied with the binominal window function is stored separately(604). A white noise correction is performed in order to reduce a spectral-dynamic range of a voice(605). A linear prediction coefficient is obtained(606), and is transformed into an LSP of less quantization error by use of the obtained linear prediction coefficient(607). A DTMF tone is detected(608). If the input signal is not a tone signal, a variable(IsTone) indicating whether the input signal is a tone signal or a voice signal, is set to a false(FALSE)(610). The LSP value is sent to a weight filter and a quantization block(614). If the input signal is a tone signal, the variable(IsTone) is set to a true(TRUE)(609), and a white noise correction is performed by use of own correlation coefficient(Rx) not multiplied with the binomial window(611). An LPC coefficient is obtained, and is sent to the weight filter(612). The LPC is transformed into the LSP(613).