| 摘要 |
PROBLEM TO BE SOLVED: To provide an echo canceller for surely carrying out decision of double-talk. SOLUTION: A sound volume ratio learning means 1 calculates a sound volume ratio of a sound output signal externally output and a sound input signal mixed with an echo component due to a sneak-passed sound output and learns a sound volume ratio of a single talk state in its own canceller. A double talk detection means 2 detects a double talk depending on whether or not the sound volume ratio calculated by the sound volume ratio learning means 1 this time is adaptive a double talk state predicted from the learning of the sound volume ratio so far and informs an echo cancel processing means 3 about the results. The echo cancel processing means 3 determines execution/non-execution of the learning operation for adaptive control depending on whether or not the double talk exists. COPYRIGHT: (C)2007,JPO&INPIT
|
