舒适噪声生成算法研究

舒适噪声生成技术的研究目的在于能够更为接近的还原实际的背景噪声.具体的实现方法是利用线性预测算法在接收端对背景噪声进行还原,即利用背景噪声的有效载荷包括噪声级别的描述以及反射系数形式谱信息,在接收端生成舒适噪声.而系统模型的阶数则不必传输,因为这个数值是可以从有效载荷的长度中获得,从而能够较好的还原出语音通信中的背景噪声.舒适噪声生成算法的评估是通过引入舒适噪声有效载荷完成的,通过比较原始语音信号以及经过静音检测、CNG算法还原出的语音信号,结果表明在无噪声背景以及有低频噪声背景的情况F,对原始语音信号经过静音检测后插人生成的舒适噪声,能较好地恢复出原始的语音信号.

Research on comfortable Noise Algorithm

The aim of the CNG algorithm is to reproduce the background noise more similar to the original ones. The core of Comfortable Noise Generation (CNG) algorithm implementation is based on the LPC algorithm in order to reproduce background noise at the receiver,namely it uses payload definition of background noise to describe the noise level and spectral information,thus to generate comfortable noise at the receiver. It is unnecessary to transmit mode order,which would be achieved from the length of payload def...