1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Poll: Audible Difference between FLAC and 320kbps MP3?

Discussion in 'Sound Science' started by el_doug, Aug 10, 2009.
4 5 6 7 8 9 10 11 12 13
15 16 17
  1. Pepsi
    There's probably is a difference between the two, but if it takes this much analysis and concentration i think it's fair to say the compare and contrast is very marginal. Like Krmathis i just go with lossless for the simple peace of mind. If i were to choose something like my portable music player, 320 would suffice.
  2. DuffyDidIt
    They sound much better in a direct comparison on my polks and sony cheap headphones headhpones.
  3. Hero Kid
    "I cannot tell a difference 99% of the time"

    For that very reason I archive all my music in FLAC wherever possible.
    However I convert all my FLAC's to LAME encoded -V0 mp3's for portable listening.
  4. JxK
    Since reading through this entire thread is daunting I'll just post a few thoughts.

    I truly think that most people, given a proper ABX will find it almost impossible to tell the differences between 320kb/s mp3 and FLAC. Of those who can, it will only be through extremely critical listening and repetition of short song segments, basically the type that is no longer enjoyable. I'll add that for portable use 320 is a waste of space. V0 VBR mp3 sounds similar enough that I doubt there are 10 people in all of headfi's thousands who could ABX it from 320, and V0 has the benefit of being 50% smaller in size.

    The only time when mp3 isn't enough is mainly with harpsichord music. I'm not sure why it is, but for some reason harpsichord just destroys mp3. Some treble in certain heavy metal songs mp3 doesn't do a great job with, but it's nothing compared to the travesty that is harpsichord. If anyone could explain why that is, it would satisfy a personal curiosity. [​IMG]
  5. nick_charles Contributor

    Originally Posted by JxK /img/forum/go_quote.gif
    The only time when mp3 isn't enough is mainly with harpsichord music. I'm not sure why it is, but for some reason harpsichord just destroys mp3. Some treble in certain heavy metal songs mp3 doesn't do a great job with, but it's nothing compared to the travesty that is harpsichord. If anyone could explain why that is, it would satisfy a personal curiosity. [​IMG]

    I took a segement of Bach's BWV 529a Concerto in G major after Sachsen-Weimar and ran it through a frequency analyzer. While most music is centred on the 1k - 4K range with **rapidly** diminishing harmonics, the segment I ran showed a smooth and gradual decline in energy from -58db at 3K to -76db at 19.2K so that when the Mp3 encoder (Lame 3.97 VBR 0) hacked it off brutally at 19.4K the FR fell off a cliff. I do not suggest that many folks would be able to consciously actually hear much of the lost frequencies, but it does remove a lot of energy from the original wave form this may itself be audible ?

    But only a DBT would show if this might be a factor. You could nudge the level up a tad to adjust for this.
  6. leeperry
    I don't have the slightest interest in lossy audio, but I always use this harpsichord song to check for middle ear resonances EQ: YouTube - The Stranglers - Golden Brown

    if it sounds good, everything will IME.
  7. JamestheRipper
    i think that its been pretty well justified but i really think it comes down to the specifics, ie artist, genre, production etc.
    i find the two biggest factors to be the genre of music and the original production.
    if im listening to relatively complex music with very good production (eg power metal), i can tell the difference, reasonably easily.
    if im listening to some techno produced by someone sitting in their room playing around with audio clips, or listening to something not particularly complex like rap or hip hop (which is basically just a synthesised beat and a little bit of singing [which has also been modified by computers to give the smoothest transition from note to note]) then i cant tell the difference.

    the only major difference for me is when im listening to a FLAC vs x<192 mp3 (im not sure in what codec though).
    my set up is westone um3x and a cowon s9. (i dont have a desktop just a laptop [sony z11wn/b] and i cant tell the difference between 320mp3 and flac when using the laptop as a source).

    i am in the process of gradually getting all of my music into flac, just on the off chance it will sound better for a specific song. hard drive space isnt really a major issue.
    the thing which i cant stand about mp3s though, is when they have been compressed poorly and the files are jumpy or have loops or they just sound awful.
  8. Aynjell Contributor
    Heh, I have all my music stored on my NAS in flac, and mp3v. The mp3 means I can stream to my 360 without twonky transcoding it every single time it's needed, and the flac is my 1:1 backup copy. [​IMG]
  9. smurfz
    It's been a while since I posted on this thread. But I've finally managed to get my gears sorted out to do my first A/B testing.
    Conclusion: I can't tell the difference. Yet..

    Source: The Four Seasons by Amandine Beyer & Gli Incogniti - 24bit 96khz FLAC
    Track: 5 Concerto Pour Violon RC372
    A: Xfi-USB -> Toslink -> DAC1
    B: USB -> DAC1
    Headphone: HD650
    The MP3 file was encoded by XLD using -q 2 down-sampled to 44.1kHz

    First, I played the same lossless file to make sure that there weren't any differences. Indeed, I couldn't hear any.
    Then I played the different files and then swapped them around a few times.

    I can pick out 192kbps MP3's fairly reliably, but only in places where there is a wide dynamic-range. Interestingly, I seem to pick out the details in the lower frequencies. In the sections where there are only violins, I can't tell the difference even with 192kbps vs lossless.
    Just out of interest out there:
    96kHz 24bit FLAC: 109MB
    320kbps MP3: 12MB
    192kbps MP3: 7MB
    I would love to hear from more people on the samples that they could hear the differences.
    Whenever time permits, I will do some more..
  10. spahn_ranch
    To my ears, and I do trust them all the way on this, the difference in detail retrieving, overall sound quality and enjoyment is most of the time greater between ALAC processed by Amarra vs ALAC by itunes, than it is between ALAC and lame 3.97b v0 by itunes.
    That's on a mid-09 MacBook Pro with 4GB RAM and a regular spinning HD. Some are also claiming, and among those the code creators, that there's more sound quality to have on a machine optimized for music use only, equipped with an SSD.
    *shrug* and FWIW,


  11. smurfz
    I don't understand how it can produce better details when:
    - FLAC/ALAC are defined codecs
    - Bit-perfect output 
    But nevertheless, I'm convinced enough to give it a go. I just signed up for a demo. :)
    Will report back if there are any revelations with my setup.
  12. spahn_ranch
    The only official explanation I've seen from Sonic Studio (Amarra) is something along the lines that there are always floating point calculations going on in digital audio, and that Amarra's calculations spit out more decimals than other players do. Which means I suppose, that a bit is a bit is a bit: not necessarily, or even simply not.
    Be it as it may with that, and I wouldn't have a clue about the science, the claim of more decimals rings true with what I'm hearing: significantly less edginess, or digititis, and more detail, which indicates it's indeed not a snake oil smearing process. Imaging, timing, openness; it's all up by a leap, plenty enough to leave any placebo for dead.
  13. aimlink Contributor

    I checked out Sonic Studio and Amara and all I could say was 'good grief.'  The price!!  Wow!  [​IMG]
  14. Head Injury

    How exactly? The illusion of more detail is probably one of the most common symptoms of snake oil.
    The easiest way to prove it's not snake oil is an ABX test. Which is very, very easy with digital files like this. If you've got Windows handy just download Foobar and the ABX plugin and load the two files up.
    smurfz likes this.
  15. smurfz
    Thanks for this. I will do my further tests with this setup.
    I've just spent an hour or so testing the demo version of Amarra Mini. Since it has on-off button where it will bypass Amarra, it's like A/B testing. But there is a bit of a gap/delay while switching, so it is not as easy to do A/B testing.
    So far, in most situations, I can't hear the difference. A very few times, I thought I heard a difference, but I couldn't reliably pick what Iiked more.
4 5 6 7 8 9 10 11 12 13
15 16 17

Share This Page