Before getting to my results, I'll mention that I tried this on my computer system (Audioengine 2 powered speakers with built-in USB DAC) and my home system (Mac Mini, Perpetual Technologies DAC connected via Toslink, Krell KAV-300 integrated amp and B&W Nautilus 805 speakers, Kimber Kable interconnects). I waited a few days between tests to forget what samples were what sampling rates - but the website appears to randomize the order of the songs and samplings anyway.
Marco says this, which I disagree with:
This is what’s so frustrating about audiophiles: they convince themselves, and others, that there’s always more sound quality to be had (and noticed) with an endless money pit of obscure components and upgrades.
And this, which I do agree with:
The most noticeable quality improvements, by far, come from better-recorded music and better speakers or headphones.On my computer system, I was able to screen out the 128 kb/s MP3 in 5 of 6 cases. 320 kb/s MP3's are indistinguishable from uncompressed WAV or AIF on this sort of system. I actually picked the MP3-320 in 3 out of 5 where I picked the higher resolution the Flip of the coin, basically.
On the home system, I got 5/6 right in eliminating the MP3-128: got 3/6 right in identifying the uncompressed WAV, picked the MP3-320 as the better in 2 cases, and picked the MP3-128 version of the Susan Vega clip.
Some conclusions, and comments:
- Hi-res MP3 (320 and above) is virtually indistinguishable from uncompressed files, even on high-end gear.
- Your odds of hearing the difference improve as the quality of your equipment does, which contradicts Marco's point. There's a limit, however, and after a certain level of spending, you're chasing the wind. I think the main difference of opinion here is how high that limit is in dollars.
- A test like this is not a good way to validate your equipment and hearing. It takes time and repeated listening to notice differences. A quick A/B test sounds "scientific" but it isn't adequate.
- But Marco's right in that well recorded music makes most of the difference in sound quality. Only a small fraction of commercially recorded music is good enough to really show off high-end equipment.
- With that in mind, the selections for this demo suck. The Katy Perry is hopelessly compressed, as is the Jay Z. The Mozart is from an ancient analog recording on CBS Masterworks, a classical label not known for their quality. Everything Neil Young does sounds like it was recorded in a bathroom. The Coldplay sounds like typical studio rock.
- I couldn't hear the difference between any of the Susan Vega versions, but that's probably because that particular song was used years ago to tune the MP3 encoding. All versions sound the same by design. Doesn't mean that MP3 will always sound as good.
- A much better demo would be to take some well recorded CD's, rip them to Apple Lossless, then rip them to MP3-128, and play them side by side on the same system. Stop by sometime and I'll play you some Patricia Barber or Pink Floyd, both resolutions. You'll hear the difference.
My results on the main system:
- Katy Perry - Dark Horse: I identified the uncompressed WAV as better. Might have been a lucky guess.
- Murray Perahia - Mozart: I picked the MP3-320. The couldn't find a Delos or DG digital classical recording to use?
- Neil Young - There's a World: I picked the MP3-320. How could you tell anyway?
- Jay Z - Tom Ford: I picked the uncompressed WAV. The main way to tell here was the bass, which really benefits from higher res (you'd think it would be in the treble, but it's always the bass that gives away the compressed versions).
- Coldplay - Speed of Sound: I picked the uncompressed WAV
- Susanne Vega - Tom's Diner: all sounded the same to me, I picked one at random and got the MP3-128.
