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

FLAC vs. 320 Mp3

Discussion in 'Sound Science' started by icedup, Sep 7, 2011.
29 30 31 32 33 34 35 36 37 38
  1. castleofargh Contributor
    I wouldn't bet that the way a file is analyzed by an encoder will secure a definitive result once it's been done once or twice. I think that for lossy formats the rule of thumb of not editing the file too many times, should stand. but maybe I'm wrong about that for at least some encoders?
    on the other hand, we do know that from an audibility standpoint, it's usually no big deal. some years back I remember people running like a hundred or more encode-decode for fun ... I mean science.
  2. bigshot
    Try it! I have. Then you'll know. I once got tired of hearing people say that playing 78s with a steel needle would wear them out, so I took a new old stock 78 and made a digital capture of it, then played it 100 times, then made another digital capture. By the end I was sick of the song. But there was no audible difference between captures. People depend too much on common knowledge and what seems logical, when a simple test can tell them the truth of the matter.
  3. stonesfan129
    I think the difference is supposed to be easier to hear with high-end source gear and good headphones. I listen on a FiiO X1 1st Gen and Sennheiser HD598SE. I cannot hear the difference between 256k AAC from iTunes and a FLAC CD rip. I don't really know what differences to listen for either. I try to buy the CD or FLAC download when I can just so that I have the best source file for archival. People with better source gear and headphones than me say the difference is easy to tell and that AAC and Opus are more transparent to the lossless source than MP3.
    TheTrace likes this.
  4. b0r0b
    I can't hear the difference on AD1000x's when not hooked up to an amp. I don't feel like it's a massive difference regardless, but I also may not have the "ears" for it. I do however have the "eyes" for 144Hz monitors, etc.
  5. RRod
    The issue with headphones at least is that the tuning of the specific cans amounts essentially to an EQ. If there's an artifact with content near 6kHz and your cans have what amounts to a massive EQ boost at 6kHz, you are more likely to hear the artifact.
  6. maxtreme
    I can't tell between 256/320, but for certain tracks with detailed highs, differences WAV and 320 are pretty audible.
  7. maxtreme
    Yes, 144hz vs 60hz to me is like WAV vs 128kbps.
  8. bigshot
    I have a test with ten particularly difficult to encode samples... Fraunhofer MP3, MP3 LAME and AAC at 192, 256 and 320 along with a lossless sample. Each codec has its own threshold of transparency, and this test is a good way to check to see where that threshold lies for you. If you would like to take the test, feel free to PM me and I will set you up. I think you might be surprised what you find with a blind test.
    Brooko likes this.
  9. kriscm
    This is my first post here on head-fi. I decided to make an account and chime in, and I’m sure I’ll be posting more in the future. I’m on mobile so excuse the auto-correct and if I did something wrong.
    I’ve spent quite a few hours examining and putting to the test various mp3 encoders.
    I once read a page online which measured the effects mp3 compression had at different bit-rates to the noise floor of a 16-bit file as well as the removal of certain frequencies the encoder figures were not needed.
    I discovered LameXP a few days ago and figured I’d look into it since it utilizes the lame 3.100.
    After much experimentation; if I set the algorithm quality to highest, used V0, joint stereo, and enabled bit-rate management to 320kbps, the 16-bit dither was also retained without much loss.
    Using experiments at lower bit-rates I was shocked to discover that V0 at 128k sounds better than CBR 128k. ABX tests were easy on that part.
    V0 has many features that CBR does not include, such as noise shaping, and when forced to stay at one bit-rate it seems to perform better than CBR.
    I never got into programming my encoders with custom commands and have always preferred using a GUI.
    I still convert my lossless to 40khz sample rate with Audacity at Best Settings and back to 44.1khz to the encoder so the high frequencies past 20khz don’t weigh down the encoder and tends to even keep a shaped dither intact and mostly artifact free.
    I recommend LameXP and the settings stated above if one wants to get the highest quality setting out of mp3.
    (I have a lot of classical and dynamic/quiet recordings)
    I don’t see the need for lossless except for archiving. I like to keep the copies in case I have to re-encode and will always consider my lossless rips to be pure gold. I love my newly encoded mp3s because of portability, compatibility, and quality.
    Putting the lossless and the mp3 together while reversing the phase of one of them highlights the info removed during the encoding and luckily the difference between the encoders and encoder settings is measurable.
    It was fun to examine, test, compare, measure, and figure out the highest quality mp3 encoder and settings, but I’m glad I have finally found the perfect setting and encoder..... until lame gets updated again :p
    Last edited: Apr 16, 2019
    RRod likes this.
  10. bigshot
    You should always use VBR. It only helps, it can’t hurt. I determine the threshold of transparency at CBR, then use the same setting with VBR for use.
    Last edited: Apr 16, 2019
  11. Lulu801
    Ah so peaky treble or bright signature could reveal flaws in a encode that would be fine on a smoother or dark headphone?. Could explain why i notice a few on my ER4SR but struggle with my SRH1540 since the shures are dark.
  12. Tsukuyomi
    I've done 320kbps and FLAC 16/44.1 testing with my own headphones and gear, and i can absolutely say with confidence that i can hear a difference.
    That being said, im not using your typical "ipod earpods" which people can argue using better headphones makes it easier to hear the difference or not.

    I'm using my favourite headphones at the moment, my DT1990 Pro and my B&O H6(2nd gen). and the source is the same, its my RME ADI-2 DAC. playing the two files on Foobar2000.
    same song. with turning away from the screen and my amp/dac. my amp dac has a remote that allows me to shuffle between songs so its absolutely random.

    every time i've made the choice on which is 320kbps and FLAC 16/44.1 i've been right. its noticeable for me. that being said, the only time i have difficulty between song quality detail is when im testing between 16bit and 24bit. im usually 70% accurate with thoes two detail points in respect.
  13. castleofargh Contributor
    1/ there is an abx component for foobar, so start there. it's better than a fake blind test. although I did sort of the same thing on DAPs for years, until I realized that I often could tell the file based on how fast it started or by the failure to play mp3 gapless.
    2/ I'll assume that you have converted the tracks yourself so you know they're not different masters, and that you don't have replay gain applied on one of the track. also that you're not using an MP3 encoder from the fifteenth century. always worth checking just in case.
    3/ if after that you still do get pretty consistent success, try lowering the volume in foobar by maybe 2 or 3dB in case what you're hearing is only intersample clipping. I personally notice it only on very few tracks, but I don't have the best listening skills in the world and I'm guessing the musical genre matters too, so it can be worth trying to be real sure it's about mp3 and not some side effect that we can all avoid.
    4/ if by now you still pass fairly consistently, I'll just accept that you know how to detect mp3 from PCM even at 320kbps. ^_^
    sonitus mirus likes this.
  14. bigshot
    Frauenhofer MP3 isn't the same as LAME, and AAC is a step above both of those. People tend to lump lossy all into one category. I haven't found anyone who can discern high data rate AAC yet. The other factor is the recording you use to compare. Some reveal flaws better than others. With some simpler to encode files, I doubt people could even discern 192 or even 128.

    One dead giveaway is that you say you are discerning "sound quality detail" and you relate it to 16 vs 24. That's apples and oranges. Detecting lossy is all about artifacting, not detail. And 16 vs 24 is all about noise floors, which have absolutely nothing to do with detail. I think you are allowing bias to creep into your tests. They may not be as blind as you'd like to believe they are.
    Last edited: May 13, 2019
  15. stonesfan129
    I try to keep my source files on my server in FLAC when I can. I have purchased single songs from Google Play which uses 320kbps CBR MP3 (I have noticed they are a mix between LAME and "Free Format" for the encoder). I thought it was going to sound bad but I am pretty happy with the sound quality. I think it's worthwhile to keep your sources in lossless as it gives the most flexibility in encoding to other formats. But if all I had were my MP3 and AAC files, it wouldn't be the end of the world. Keep in mind there are different MP3 and AAC encoders. LAME is considered the best encoder for MP3. Apple/iTunes is considered the best encoder for AAC.
    Last edited: Sep 14, 2019
29 30 31 32 33 34 35 36 37 38

Share This Page