A VOICED/UNVOICED DECISION METHOD FOR THE SMV OF 3GPP2 USING GAUSSIAN MIXTURE MODEL

摘要

PURPOSE: A real time voiced/unvoiced sound classification method for SMV of 3GPP2 using a Gaussian mixture model is provided to extract a characteristic vector with superior performance of voiced/unvoiced classification among existing characteristic vectors of SMV and to apply the extracted characteristic vector to a characteristic vector of GMM in order to classify voiced/unvoiced sounds, thereby improving voiced/unvoiced sound classification performance while minimizing additional computation amounts. CONSTITUTION: A real time voiced/unvoiced sound classification method for SMV of 3GPP2 using a Gaussian mixture model comprises the following steps: a step of extracting a characteristic vector with superior performance of voiced/unvoiced classification among characteristic vectors of SMV(S10); a step of classifying voiced/unvoiced sounds by applying the extracted characteristic vector to a characteristic vector of GMM(S20).