Audio Compression using wavelets in MATLAB
Algorithms compared:
Haar Algorithms
Daubenches Algorithms
Parameters of Comparison
PSNR : Peak signal-to-noise ratio
NRMSE: Normalised root-mean-square error
Compression ratios
MATLAB Inbuilt functions used:
audioread
Audiowrite
length
ceil
Sqrt
DCT
IDCT
wavedec
ddencmp
wdencmp
fft
compand
waverec
Concepts involved: Discrete Cosine Transform Inverse Discrete Cosine Transform Windowing
Results: Haar Algorithm
PSNR : 55.4912 MSE: 0.4285 Compression Ratio: 1.9998
Daubenches Algorithm
SNR : 69.7492 MSE: 0.0830 Compression Ratio: 2.1594
Conclusion:
1:After analysing the observations we see that the compression ratio of the haar algorithm is less than daubenches algorithm so the audio compressed by daubenche’s algorithm will require lesser space than that by haar’s algorithm.
2:Daubenches algorithm is best suited for lossless compression of speech signals as it has more PSNR and substantially low NMSE.