Voice segment determination program speech section determination device, and the speech segment determination method

摘要

Update the parameters to be used for the voice segment determination without applying a burden on the user, Robust speech section determination device to noise, to provide a voice segment determination program and speech interval determination method. By comparing with a threshold value of the feature time series of the input sound, non-speech section (first speech segment) speech segment time series of the input sound speech segment determination device, (first non-speech section The feature values ​​of the time series of the first non-speech interval after superimposing the speech section first determination unit, a standard audio obtained from the standard voice storage means in the time series of the first non-speech period determined and) By comparing the values, and the threshold value, the speech section the second determining section, and the second to determine the non-speech interval and the speech section of the time series of the first non-speech interval after superimposing the standard voice I comprises a threshold updating means for updating the threshold mismatch ratio and the determination result of the speech segment determination means 2, the correct calculated from the standard voice becomes smaller.