摘要 |
PROBLEM TO BE SOLVED: To automatically extract only a desired signal from the desired signal mixed with musical sound and noise by calculating an estimated value valuating an extent of desired signal extraction from a ratio of obtained basic frequencies and the ratio of obtained short time power. SOLUTION: The data of respective constant lengths are cut out in voice input parts 1, 2 by taking out the synchronism from an input signal line starting from an input signal and an output signal line corresponding to that. Respective cut out frames are sent to a short time power processing part 4 of a next stage, and the short time power of both frames are calculated, and the ratio Rpow of the shot time power of these input/output frames is obtained. Further, amplitude envelopes are obtained from the data of input/output buffers in a modulation processing part 5, and respective amplitude envelopes are Fourier transformed, and the basic frequencies are extracted. The ratio Re of the basic frequencies is obtained from the basic frequencies of these input/outputs. An estimated value calculation part 6 calculates the estimated value valuating the extent of the desired signal extraction from the ratio RPOW of the short time power and the ratio Re of the basic frequencies. |