Thanks, might I ask your reasoning behind that? I've read different answers about it, but I forget where. There was a great ars technica (?) piece about it, but maybe it was over my head since I came away without a definite conclusion. I'm pretty sure someone who frequents this cafe knows for sure.
No worries. Data compression does not concern itself with dynamic range. It concerns itself with file size and attempts to preserve the fidelity as much as possible. In the case of lossless compression, full fidelity is preserved. Most of the lossless approaches use some sort of entropy coding. One of the best known ones is Huffman. A more advanced one is Arithmetic. All of these methods (lossless) look at data statistics to compress BITS. I'm fairly familiar with lossless data compression. Lossy uses filter banks, transforms (cosine for example), and quantizers (such as vector ones.) These will lose some information though. Lossy compression is many times used combined with lossless compression.
I'm not as familiar with audio dynamic range compression, but my understanding is that it applies a gain mapping where the instantaneous gain is mapped to a different gain using a linear (or quasi-linear) function. My yamaha receiver calls this silent cinema I think. This does not necessarily translate into a smaller file. We are talking dynamic range compression vs. file size compression.
Edited by ultrabike - 12/4/12 at 6:36pm