摘要 |
PURPOSE:To effectively process only an input voice data at the sound period into a packet by setting a threshold value to a proper value in response to the fluctuation of an inputted voice data and noise level. CONSTITUTION:The mean power E of a voice data in a digitized voice signal is calculated by a mean power calculation section 2 and the threshold value L of the mean power E is revised sequentially and compared with a threshold value L of a threshold value storage register 6 by a comparator 3. In case of E>L, the data is discriminated to be a data block of the sound period and in case of E<L, it is discriminated to be a data block of no sound period and only in the sound period, the data is processed into a packet by a packet composing section 4 and sent. On the other hand, a power calculation section 7 calculates the mean value of the mean power E of the block discriminated to be the no sound period over a prescribed period at the no sound and gives a value higher than the value by several dB to the register 6 and an inter-block power ratio calculating section 8 sets the threshold value L when the power is changed comparatively rapidly.
|