摘要 |
<P>PROBLEM TO BE SOLVED: To provide an audio signal processing apparatus capable of excellently separating audio signals of a plurality of sound sources from two systems of audio signals including audio signals of a plurality of sound sources. <P>SOLUTION: Two systems of input audio time-series signals are converted into frequency domain signals, respectively. Regarding two systems of frequency domain signals, a level ratio or a level difference between corresponding frequency division spectrums is calculated. Frequency components with the calculated level ratio or level difference being a predetermined value or near the predetermined value, are extracted by a first sound source separating means and a second sound source separating means and the extracted frequency domain signals are converted into time-series signals as first and second output audio signals. The frequency domain signals from the first and second sound source separating means are subtracted from frequency domain signals of input audio of two systems, and frequency domain signals of the subtraction results are converted into time-series signals as third and fourth output audio signals. <P>COPYRIGHT: (C)2008,JPO&INPIT |