摘要 |
PURPOSE: To reduce occurrence of abnormal sound caused by conversing an ADPCM code having no error and to improve voice quality in auditory sense by utilizing correlation between an ADPCM code of the present frame and an ADPCM code of the past. CONSTITUTION: Correlation between an ADPCM code of the present frame and an ADPCM code of the past is obtained by using the ADPCM code of the past, a rate of error's in a frame is decided by a matching processor 27 and an error detector 23, and conversion of the ADPCM code is performed by an ADPCM code convertor 24 based on the decision result. Converting an ADPCM code having no error is made less by deciding a rate of errors in a frame, also occurrence of abnormal sound having large amplitude of a decoded signal can be prevented, voice quality in auditory sense at the time of voice output can be improved. |