Research And Implement Of Digital Audio Restoration Based On Wavelet Transform

Academy of Music and archives across the country have accumulated of a large number of historical audio materials, which have special values of history and culture and at the same time are the only existing copies. They are non-reproducible. As time passes and the technical limitations of carrier material, some of the early audio materials are on the verge of disappearing and the current situation of preservation is worrisome.With the rapid development of computer technology, computer based information processing plays an increasingly important role, and digital audio processing technology has made rapid development as well. In order to better make use of the valuable historical audio to serve future teaching, research and commercial applications, historical audio material needs to be digitized, but overall noise and localized noise often appear during the process. Digitized audio restoration is necessary in order to try to get back to its appearance, while keeping the original audio signal to its largest extent.This paper takes a key scientific and technological project of Shanghai Municipal as background, which is the research of digital restoration and preservation of digital of valuable historical materials. The major works involved in this paper are as following:(1) It focuses on the one way to reduce Hiss Noise—wavelet thresholding, with the help of simulation experiment, it discusses the influence different wavelets, threshold Functions and threshold select rules would have on the noises reduction performance. The research result has a certain reference value.(2) An adaptive multi-thresholding for audio denoising based on wavelet packet transform is proposed. Threshold value of this method is localized and superior to overall threshold. The thresholding is derived in a Bayesian framework, which is near-optimal in the sense of mean square error. The experiment shows: this method could save as much as original audio signal, and has significant effects on HISS noises elimination. It overcomes the general the shortcomings of most wavelet threshold have, over- eliminate the wavelet coefficient.(3) The paper proposes one solution to meet the needs of historical digital audio materials preservation—MemoryAudio. MemoryAudio concludes the function based on the adaptive multi-thresholding for audio denoising based on wavelet packet transform, the bulk repair functions, the multi-versions functions etc, which overcomes the general shortcomings of most audio processing soft wares that they can not save the audio restoration information and assessment information. The audio restoration modules adopts plug-in technology, makes it possible to extend its functionality.This paper uses digital signal processing method, re-process the audio signal and lower the complexity of the audio restoration of the past, and makes valuable audio materials on the biggest play to its historical value and cultural value. Meanwhile, the audio restoration modules adopts plug-in technology, makes it possible to extend its functionality by adding plug-in according to different needs of restoration staff or professional staff. That is more flexible and agile and improves the efficiency of audio restoration.