1. On the credentials, you got me. Apologies.
2. I've read that paper. From what I understand, it tested HE-AAC at 160kbps. This is still much different than "basic" AAC at 256kbps, which is the standard I would consider the norm for modern lossy. It has been the standard for iTunes for a few years now. These are important variables to define and control in this discussion.
Heck, if we did a test with 96kbps, my deaf old ears could pass it, something I would happily place money on.
First off. No worries. Sorry I got defensive on the credentials thing. I've had a horrible morning and my patience is a bit on the thin side. Also, I know there's a lot of trolling and smoke-blowing on this site sometimes so I totally understand being skeptical of someone that would make my claim.
Secondly, Wow! you read the paper?
Yes the tests were done at 160kbps. We were testing broadcast standards which are or at least were to that specification. However, we did tests beyond the scope of the paper at higher bit rates and found similar problems. The problems were obviously magnified at lower bit rates though.
The crux of the biscuit (Zappa reference) is this: Any time you remove information, you add noise and distortion. Plain and simple. It can be drastic or subtle but it's a change in the audio.
The best example I can think to give is this:
I work as a freelance engineer at a studio called Immersive Studios.
Immersive
We have a DSD studio. DSD stands for Direct Stream Digital. It's one bit at 2.8224 MHz sample rate. It's the closest thing to pure linear sound that you can get in a digital format.
When we record things in DSD the audio is clean, deep, and amazingly transparent. As soon as we dump it to Pro Tools, even at 96kHz 24 bits, it sounds like the image flattens out and the transparency disappears. It's because of the quantization errors inherent in PCM audio. Take that 96kHz 24 bit PCM and dump it to CD quality 44.1kHz 16bit PCM and even more depth and transparency disappear. Then convert it to AAC or MP3 and more and more and more and more information goes away and the more crap gets added.
The best analogy I can think of is DSD is like having the original Mona Lisa in your hands. PCM is like have a pretty good photographic copy. 96kHz would be like a nice reproduction on good quality paper, and 44.1 would be like a decent inkjet print. AAC and MP3's are starting to get into the world of a fax print. Sure you can see the image, but it's pretty grainy and full of errors.
The nice thing about digital audio is we can all own an original Mona Lisa. As in, we can all go out and buy an SACD (the format that plays DSD audio) and have an exact duplicate of the pure original source.
Anyway, that's how I see it from the experiences I've had. I'm by no means the be all end all authority on the subject. Far from it. But I thought I'd share since I think I have a little more experience on the topic than most people.
