A friend is trying to tell me that "lossless is pretty much a lie," when it comes to audio. He converted some song into "Apple Lossless," and swears he hears a difference between that and the AIFF. This is something he heard from an audio-expert friend. This was basically a sub-debate of an analog vs. digital debate.

I know that lossless compression works, since zipped applications still execute when they're uncompressed, and .png compresses images and is demonstrably perfect. My Mechanical Engineering background only gives me a basic understanding of signals as it is, which helped with understanding audio. The problem is that I don't know enough about compression theory to debate this.

Can anyone link me to information that could disprove this? Or is it true from some point of view? Thanks a lot.

My suggestion is reading through the Lossy Format Comparison thread right above this one.

I've read the thread, I understand lossless compression is possible, I can't imagine why it wouldn't apply to audio. He swears there's a difference. But lossless, by definition, will decompress to the exact same file that was compressed.

Is Apple's lossless format not truly lossless? Maybe he encoded it to a lossy format by accident (I think ALAC uses the same file extension as the lossy AAC)? I'd tend to lean toward the "he's hearing things" option, but that just amounts to blowing him off.

I'm thinking of ripping something and comparing the waves somehow.

Lossless compression of an audio track means that all the data from the CD is in the track, and the exact sound from the CD could be replicated back from the data. However, the data itself is still limited by the CD's bitrate in the first place and so compared to an analogue format, one could consider data to be "lost"...

But a lossless rip should sound exactly like the CD. Maybe when your friend encoded the file, he unwittingly applied some kind of replay gain which could have distorted the sound?

Yeah the more I look into it the more it looks like he's hearing things or compressed it wrong. I had read that some lossless image formats have to drop the number of colors to 256 if it's above that and then dither it, but CD sampling rate is 1411k, and there would be no reason for ALAC to drop it.

Sorry for thread revival, but this is interesting.



CD sampling rate is 44.1 khz, bit depth is 16 bits. You're talking about bit rate there.

I bounced a mix from Pro Tools, and converted it to Apple Lossless. Then I uncompressed it, imported it into a new Pro Tools project with the original AIFF, each to their own track. I phase inverted one, so that all the differences between the two would immediately be audible. There weren't any.

Lossless is lossless. It's all in his head.

Best way to test:

Rip a file uncompressed, then encode it to a lossless format that you KNOW isn't touching the sound (as in, not actually lossy, or replay gain, or anything).

Then decode it again, and compare it to the original file. If it's even a bit off, then one could claim there is loss.

Or do what Big T said. Yay for not reading!