An effective storage method during a sampling of speech signals

It is necessary for speech samples to be stored in a memory buffer before they are processed by speech analyzers without a real-time processor. An algorithm is proposed that uses the buffer efficiently when analog speech signals are converted to digital samples by the analog-to-digital converter. In order to implement this method in real time, the buffer is divided into the starting buffer and the remaining buffer. Until a voiced speech is found, the converted samples are sequentially stored in the starting buffer. The buffer is then shifted. When a voiced speech is found, the next samples are sequentially recorded in the remaining buffer