摘要 |
PROBLEM TO BE SOLVED: To solve the problem that allophone is generated and causes discomfort to a listener when a bit stream of erroneous MPEG audio is reproduced by an MPEG audio decoder. SOLUTION: The speech decoder comprises an error detection part for inputting data compressed in frame units and detecting errors in frame units, a synthesis part for expanding and synthesizing the compressed data, and a silent frame generation part for generating a silent frame, and when the error detection part detects an errors, the silent frame generated by the silent frame generation part is used in the synthesis part to synthesize the sound. |