摘要
采用短时谱分析、合成技术 ,对含噪语音进行帧间重叠谱减法消除噪音 ,这种算法符合语音特性连续变化的特点。实验证明 ,该方法有效去除了噪声干扰 ,得到了增强语音 。
This paper uses the time-varied spectral analysis and the synthesis technology eliminate the noise from the noisy speech through the frame-overlapping-spectral subtraction method which accords with the characteristic of constant changing speech. The test verifies that this method can effectively eliminate the disturbance of noise to get the enhanced speech, and ensure the non-loss of the intelligibility and the naturalness of speech.
出处
《空军工程大学学报（自然科学版）》
CSCD
2001年第5期48-50,54,共4页
Journal of Air Force Engineering University(Natural Science Edition)
基金
空军科研部基金资助 (KJ0 0 10 4A)
关键词
时变谱
谱减法
语音增强
帧间重叠
time-varied spectrum
spectral subtraction method
speech enhancement
