Abstract: In forensic acoustics,noise reduction is mainly applied for two purposes;one is to reduce noise mask effect and make speech more intelligible,and the other is to diminish noise interference and improve sonogram quality.In the experiment, more than 20 kinds of noises were recorded;then each was synthesized with speech into 4 samples with different SNR levels.The samples were processed using three noise reduction tools.It was found that mask effect could be effectively reduced for most samples and sonogram quality could be improved for some samples. The effect of noise reduction on speech was also discussed.

Key words: forensic acoustics, noise reduction, voice identification

摘要： 司法声学中有两种导向的降噪处理模式:一是在语音内容辨听鉴定中,通过降噪处理,减弱噪声的掩蔽效应,提高语音听觉的清晰度和可懂度:另一种是在说话人鉴定中,通过降噪处理,减弱噪声对语音声学特征的干扰,使语图更为清晰可辨。本实验中,录制了日常生活中经常遇到的20余种噪声。通过混音,对每一个噪声合成出四个等级信噪比的含噪语音声样。然后,利用STC降噪系统、Adobe Audition软件和VS6.0语音工作站,分别对含噪语音进行降噪处理。实验发现,经过降噪处理,多数声样的噪声掩蔽效应能够减弱,部分声样的宽带语图得到改善。最后讨论了降噪处理对语音的影响及利用降噪语音进行说话人鉴定的注意事项。

关键词: 司法声学, 降噪处理, 说话人鉴定