Hacker News Reader

Comment by shawabawa3 | Hacker News Reader

They are 320kbps mp3's. The difference between that and flac is practically undetectable to the human ear[0]

  [0] http://www.music.mcgill.ca/~hockman/documents/Pras_presentation2009.pdf

edit: The knocking one is 320kbps, the others seems to be lower. I imagine the pure tone ones compress very well.

View on HN

Yes. What my concern about this is is actually the lower and higher frequencies in the mp3s. With these the loss was neglectable. I thought preserving the full range and quality was paramount with audio testing, but I'm not an expert or an audiophile, it was just my preconception. (Listening to an original .wav song with my headphones and then an mp3 of 192 or 320 kpbs quality, the difference seems obvious, but mostly at the low- and the high-ends.)

You can tell the difference between 320kbps and a wav file across a wide range of music? You might want to try and get in contact with a competent researcher in this area as they may be interested in studying you

Hehe, absolutely no. A 320 kpbs MP3 and .wav are hard to distinguish, but the high freq range is usually the one where things are missing, and the lows also but only if there are a lot deep sweeps and bass. If I wouldn't have a wav and an mp3 for comparing I wouldn't be able tell the difference. I don't have the high-end headcans or the finesse for that.

Have you considered doing a test of this? Write a program that plays a random choice from the three formats, and keep track (hidden from you) of which is which. Have it quiz you on which it is playing / has played, and see if you get interesting results! :)

you could use Foobar's blind test component for this http://www.foobar2000.org/components/view/foo_abx

https://itunes.apple.com/us/app/abxtester/id427554135?mt=12 works great. Turns out I can barely tell the difference between 96k and 128k.

>The difference between that and flac is practically undetectable to the human ear

For most things, yes. There are some sounds that do not compress well, however. Harpsichords and trumpets in particular are pretty noticeable.

This is probably a failure of the major algorithms rather than a damnation of audio compression in general. I imagine even they will someday be basically imperceptible in difference.

Right now, though, I can score 100% on a blind ABx test on most horn instruments =\

Reply on news.ycombinator.com