5G iPod: I hear an improvement with AIFF versus Apple Lossless

I didn't make any tests, but since both formats are lossless, they should both sound exactly the same.

The only difference is one takes up more space than the other, and it might take longer to load the song, too (My old Rockboxed iPod had a hard time loading flac files from the hard drive fast enough).

Set up a blind ABX test.

• Make a folder or playlist or something with the same file ripped as AIFF and Apple Lossless. The file should preferably have a good point of comparison right at the beginning, so you don't have to shift around to find one every time
• Have a friend hold the iPod while you listen with your eyes closed
• Have him assign the AIFF or the Apple Lossless to "A", and the other to "B". This always stays the same, even between rounds
• Every "round" have him choose one of the files to be "X" at random. This stays the same within the round, but can change every round
• Ask him to play "A", "B", or "X" at any time, as many times as you want
• Try to determine whether "A" or "B" is "X"
• Have the friend record which one "X" really was, and your guess
• You should do at least 10 rounds, preferably more but that could get time-consuming

This will determine whether or not you can really hear a difference without knowing which file is which. I can't think of a reason the files would sound different unless there's something wrong with how you're encoding ALAC. Make sure you turn off any sound effects or EQ to be safe.

Take a line-out from the ipod - connect to the input of a sound card. Record samples from both types (for same track) using Audacity or similar. Trim and align the samples to start at the exact same place and be the same length - exactly. Run each sample thru the spectrum analyzer, save the results to text files, import into a spreadsheet program and compare.

Any significant frequency or amplitude difference beyond normal ramdom quantization error will be apparent.

For rigor use 10 copies of each sample and aggregate the results to smooth out random error.

If, and only if you find significant variations investigate further

It may be that one decoder alters the gain marginally, I have a media streamer that (verifiably) does this !

Even a slight increase in volume can be misinterpreted as fundamental changes

...or don't and accept it as a psychological artifact, which it most likely is, and go on with your AIFF madness