I keep reading about how quite a few people here listen to lossless files on their hifi stereo systems, as if listening via stereo & speakers were more critical or revealing of the source music than listening via headphones/computer. I always thought the conventional wisdom was that it's harder to discern lossy music through speakers, as opposed to headphones. So I'd like to try ABX tests of the same lossy files I listen to on the computer (itunesAAC 128vbr, which I fail every computer/headphone abx), but on my stereo with my "good" speakers. I just need some help in making it double blind Also I'm hoping to convince myself I really don't need lossless, since I'm about to buy some gear that will let me stream music wirelessly from the computer into the stereo. But if I do pass too many tests, or pass anything too easily, I won't hesitate to switch to lossless (or at least increase the bitrate).

Does anyone have any tips or links on specifically how to set up this ABX test in a non-computer environment? AFAIK, if my wife were to control which songs were being played, that would only make it single blind, correct? As source, I could either burn CDRs with decompressed WAVs and original WAVs, or use the ipod with both lossless & lossy versions. But I'm not sure how to name/tag the tracks so that nobody knows which file is playing until after the tests have been finished.

fileABX or iPod-ABX.
These two tool can randomize the order of samples for double blind test. I do recommend to make the reference and lossy equally loud (by WavGain or something) before processed by these.

I wonder if comparative tests have been done on HA about judging lossy codecs on headphones and speakers.
When public listening tests are done, is one required to specify listening conditions ?
It's my impression that the optimization of codecs is largely based on feedback from headphone listeners (at least on HA).
Are there any statistics about that and if so, does it matter ?

Although I like headphones in some situations, I wouldn't use them to make important sonical judgements during mixing or mastering.
Frequency response of headphones (examples) is often far from "flat", which can have quite some influence on masking properties.

If you can't feed your hifi stereo system with the sound from your pc, putting various samples on a cd would seem a workable solution.
You could do an ABX test by using 3 samples of each fragment: original, coded version and X (one of the previous ones).
One file will be used twice, which costs more discspace, but it makes testing easier IMO.
You'll need someone (or a tool) to put the files in random order to make the test double blind.
Please do report your findings. It's an interesting issue.

Be carfeul with WavGain or similar because it is not 100% accurate and can even make ajustments which would make different volume between lossy and reference.

I forgot to mention that when you use only wav files in your test, you bypass the decoder in your iPod (or whatever playback system). If you think the quality of the decoder has a major influence on the sound quality, it's probably better to use the encoded audio in the test. An audio cd wouldn't be an option then. You could play back files from your iPod but it'll have to be double blind (haregoo recommended iPod-ABX).

It matters for those that interpret results whether speakers or headphones were used in test. If you are listening your music with headphones then more important to you are the results when headphones are used.
Also when developing objective measurement of audio quality (e.g. PEAQ) results of subjective listening tests are used. And then it is desirable to know whether speakers or headphones were used.
I believe that headphones are better for testing, but I didn't find any proof of that so far.

Thanks, I'll try one of these. Hopefully I'll do this today if I can spare the time.

Kees, I would tend to think the iPod decoder would not be an issue on the sound, but it's a good point and I may be wrong. But regardless, the goal of my testing is just on the effects of itunes AAC at 128 kbps vbr on my stereo. So I would like to bypass the decoder anyway. And once I buy some streaming hardware, the files would be decoded by that DAC and not the iPod.