The goal of the project was the creation of a simple audio encoder-decoder system. It was the main assignment for the Multimedia Systems and Virtual Reality course of the Electrical and Computer Engineering Department of the Aristotle University of Thessaloniki.
The encoding procedure is implemented using the following steps:
- Downsample the signal
- Divide the signal into non overlapping frames
- For each frame:
- Calculate the parameters of the optimal linear predictor.
- Quantize the signal and the parameters using a uniform quantizer.
- Apply the A-DPCM.
- Encode the current window using Huffman Coding
- Write the number of bits required, the Huffman code, the first two levels of the quantizer, the minimum and maximum value of the predictor and finally the encoded window to a file. These bits contain all the information needed to reconstruct the signal.
Iterate over the windows and decode them one at a time:
- Extract the stored data from the bit stream
- Reconstruct the quantizer
- Decode the data using the huffman tree
- Apply the inverse of the A-DPCM algorithm
- Concatenate the reconstructed windows
- Upsample the signal to it's original frequency
