FIELD: data processing.
SUBSTANCE: invention relates to means for obtaining a frame of modification of speech activity. First result of the decision on the detection of speech activity and the second result of the decision on the detection of speech activity are obtained. Number of tightening frames for active sound is obtained and number of updates of background noise is obtained. Number of modified frames for the active sound is then calculated in accordance with the first result of the decision on the detection of speech activity, the number of updates of the background noise and the number of tightening frames for the active sound. Finally, the result of the decision on the detection of speech activity for the current frame is calculated in accordance with the number of modified frames for the active sound and the second result of the decision on the detection of speech activity.
EFFECT: high efficiency of processing speech signals.
20 cl, 8 dwg
