Using machine learning to determine the best wavelet for compressing acoustic data

This research uses acoustic data to determine the best method to find the optimal wavelet to use for data compression. The power spectral densities before and after wavelet decomposition, compression, and decomposition (decompression) are compared to assess the quality of the compression. Machine learning methods are tested to choose the optimal wavelet with the best qualities. Ultimately, this research will create a program to automatically give the user the best wavelet(s) to compress a particular acoustic data set. [This research is funded by the University of New Orleans Office of Research.]