Noise Cancellation mittels Wavelets unter PD/MATLAB

Using wavelets, a noise cancellation technique has to be developed. The signal which is to be denoised, is transformed into the wavelet domain, using the wavelet packet transform. An appropriate base tree for best representing the signal has to be found. Simple Thresholding algorithms are used to suppress the interfering signal in wavelet domain. Because of dealing with semi stationary intereference signals, setting the thresholds has to work adaptive using Voice Activity Detection (VAD). Therefore, the noise / interference level has to be estimated during non-speech periods. The algorithms are implemented in MATLAB.