能動音響制御と周波数スペクトル平坦化による低騒音な音声マスキング手法

抄録

Sound masking is a method that used for protecting speech privacy of conversation at public open space such as a pharmacy or a bank window. In a conventional masking, a masking sound is used to cover contents of conversation. This masking sound which is generated from a loudspeaker allow a third party not to understand what is said, but there is a problem that the sound level is increased because of a masking sound. Against this problem, we proposed Low-noise sound masking method using active noise control (ANC). The approach of this method is to make spectrum of phonemes flat by control sound. It is difficult to recognize phonemes that are processed by this method because the major factor determining the phonemes in speech recognition is the relative relationship of the peaks of its spectrum. In this paper, the sound attenuation performance of the proposed method is validated by control experiments in an anechoic chamber. Moreover, a listening experiment is carried out to investigate the sound masking performance. The result show that comparing with the conventional sound masking method, the proposed method can achieve the sound masking effect with a smaller sound pressure level.