VOICE SIGNAL PROCESSING DEVICE, METHOD, AND PROGRAM

PROBLEM TO BE SOLVED: To provide a voice signal processing device with which it is possible to suppress background noise and a voice signal from a remote speaker with high accuracy.SOLUTION: The voice signal processing device comprises: means for dividing an input signal by a time section and generating a plurality of section signals; means for generating, for each of the plurality of section signals, an indication signal that indicates whether the section is a first section that includes the voice component of a near speaker or a second section that does not include; means for converting each of the plurality of section signals into a frequency domain first signal; means for dividing each of the first signals into a plurality of frequency bands, and adjusting the signal level of each frequency band of the first signal on the basis of a noise component in each frequency band and generating a second signal; means for dividing the second signal into a plurality of frequency bands, determining a weighting coefficient of each frequency band of the second signal on the basis of the indication signal and the first signal, and weighting the signal level of each frequency band of the second signal with the determined weighting coefficient and thereby generating a third signal; and means for converting the third signal into a time domain signal.SELECTED DRAWING: Figure 1 【課題】背景雑音及び遠隔話者からの音声信号を精度良く抑圧できる音声信号処理装置を提供する。【解決手段】音声信号処理装置は、入力信号を時間区間で分割して複数の区間信号を生成する手段と、複数の区間信号それぞれについて近接話者の音声成分を含む第１区間であるか含まない第２区間であるかを示す表示信号を生成する手段と、複数の区間信号それぞれを周波数領域の第１信号に変換する手段と、各第１信号を複数の周波数帯域に分割し、各周波数帯域における雑音成分に基づき当該第１信号の各周波数帯域の信号レベルを調整して第２信号を生成する手段と、第２信号を複数の周波数帯域に分割し、第２信号の各周波数帯域の重み係数を、表示信号及び第１信号に基づき決定し、第２信号の各周波数帯域の信号レベルを決定した重み係数で重み付けすることで第３信号を生成する手段と、第３信号を時間領域の信号に変換する手段と、を備えている。【選択図】図１