FIELD: data processing.
SUBSTANCE: invention relates to voice activity detection (VAD). To achieve the technical result, the method includes: at least one feature of a first class in a first feature category, at least one feature of a second class in a second feature category and at least two existing VAD judgement results, wherein the feature of the first class and the feature of the second class are both features used for VAD detection (S102); and conducting VAD according to the feature of the first class, the feature of the second class and the at least two existing VAD judgement results so as to obtain a combined VAD judgement result (S104). By means of the technical solution, the technical problems that the detection of a VAD solution is inaccurate are solved, thereby improving the accuracy of VAD, and thus the user experience is improved.
EFFECT: higher accuracy of detecting speech inactivity frames before and after voice segments.
10 cl, 4 dwg
