| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a sound collection device automatically performing phase correction to achieve an improved sound collection capability even in the case that there is an error in a target sound estimation direction, a microphone interval, or a speed of sound. <P>SOLUTION: A sound collection device 1 has: a frequency analysis part 100 performing frequency analysis to a multichannel signal; a noise estimation part 101 estimating a noise spectrum of a reference channel; a steering vector calculation part 102 calculating a steering vector based on a target sound estimation direction indicating an estimation direction from a sound source of a target sound of each microphone, a microphone interval indicating an interval between a reference microphone and the respective other microphones, and a temperature; and a phase correction spectrum synthesis part 103 performing phase correction of each channel by using a spectrum of each channel outputted from the frequency analysis part 100, the noise spectrum of the reference channel outputted from the noise estimation part 101, and the steering vector outputted from the steering vector calculation part 102, and then synthesizing the spectra. <P>COPYRIGHT: (C)2013,JPO&INPIT |
