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

64 vs. 128 vs/ 256 MP3 listening test in Cognitive Daily

Discussion in 'Sound Science' started by steve999, Sep 10, 2008.
1 2 3 4 5 6
  1. Steve999
    Thanks. And thanks for the CD recommendations! I'm experimenting. Learning new things. That combined with my streak of silliness can lead to unpredictable results. But I'll get down to business soon enough. :upside_down:

    Last edited: Jun 9, 2018
  2. TheSonicTruth

    bigshot: by "encode in Fraunhofer" do you mean, in the original mpeg Layer 3 codec, vs LAME, etc? Where does one get that codec?
  3. Steve999
    First try:

    foo_abx 2.0.4 report
    foobar2000 v1.4 beta 17
    2018-06-10 02:03:37

    File A: 15 My Funny Valentine.m4a
    SHA1: ff9bd734689096c80f4eedf009f0afce728cd495
    File B: 15 My Funny Valentine.mp3
    SHA1: 80437ecc31224ab78b0ab5e8993400fe50672e22

    DS : Primary Sound Driver
    Crossfading: NO

    02:03:37 : Test started.
    02:04:13 : 01/01
    02:04:36 : 02/02
    02:05:06 : 03/03
    02:05:25 : 04/04
    02:05:46 : 05/05
    02:06:03 : 06/06
    02:06:15 : 07/07
    02:06:33 : 08/08
    02:06:51 : 09/09
    02:07:08 : 10/10
    02:07:25 : 11/11
    02:07:42 : 12/12
    02:07:55 : 13/13
    02:08:08 : 14/14
    02:08:20 : 15/15
    02:08:42 : 16/16
    02:08:42 : Test finished.

    Total: 16/16
    Probability that you were guessing: 0.0%

    -- signature --
  4. Steve999
    I think it's the MP3 decoder in Itunes. I can step you through everything in getting the ABX comparomoter set up in Foobar2000 if need be. It's really easy actually. I was surprised. The ABX comparometer does all of the work for you.

  5. bigshot
    What are you doing Steve? Why are you doing foobar tests? Do you understand what I was telling you when I explained how to recreate my observations? You don't appear to. I'm happy to explain clearly, but you have to listen.
    Last edited: Jun 10, 2018
  6. Steve999
    I get what you’re saying. I did all along. I’m working my way up my own learning curve. I’m like hearding cats. But I do appreciate you help. Let me get some confidence that I am not just wasting my time and fooling myself. And I was anxious to try out the comparomoter with my new Foobar2000 too.
    Last edited: Jun 10, 2018
  7. bigshot
    There is absolutely no question that you can hear the difference between a low bitrate lossy track and a high bitrate one. And you didn't even let us know what you were comparing there. It's frustrating when I'm asked for help and then the person asking wanders off into a corner and does something completely different.
  8. Steve999
    Fair enough. I did not know that what I was doing was not constructive. You know what I don’t know and I don’t know what I don’t know. Now I know some things I didn’t know that you did know and I shall proceed hencewith. :)

    Last edited: Jun 10, 2018
  9. Glmoneydawg
    Chess Records....too much to list but for blues or r and b these guys had it going on:)
  10. TheSonicTruth

    In my case that would be two rips of the same exact track off the same exact CD, one encoded at 128kbps and the other at 256, same settings(CBR, full stereo, etc). Listened to and compared via the same listening chain. Common sense I guess! Not an extreme audible difference - like comparing 64kbps and 320 MP3, or comparing 128 to WAV - but a definite one. No need to ABX under those conditions.

    Invalid listening tests:

    -Listening to the lower bitrate file through headphones, and the higher one through full-size speakers.

    -Comparing MP3 of one bitrate downloaded from somewhere(Ewww!) to one you ripped from a CD at another bitrate.

    -Comparing a MP3 rip from original CD at any bitrate to a MP3 rip from that CD's remastered version.

    Last edited: Jun 10, 2018
  11. Steve999
    In my view there is a need to test that for any one individual to make that claim. I'll grant you frau 96 CBR, but it still sounds nice to me if I am not ABXing. Other people here appear to feel that for 128 cbr frau it is more of a consensus that it is obviously audible at most times. That's fair, but I am not sure I agree. I will say it's reasonable, but for me, after all of the "prevailing wisdom" I have read on the net, surprisingly challenging. I find 128 kbps CBR frau versus lossless highly challenging for an unexceptional passage of music, say what you will about my hearing. I have not trained for artifacts before the last couple of days, but I'll say, if you are talking about day-to-day sound quality, I think you'll be surprised how little the difference is, and you might ask yourself if it's meaningful from an enjoyment perspective. At least I do. And training for artifacts might distract you during enjoyment listening.

    There's really no practical need for you to gain this knowledge--you can just rip at higher bitrates and relax, but I find it interesting to get personal knowledge of the actual differences and I do challenge you (softly) to try out some ABXing at 128 CBR frau. . . having in mind that is a worst case real world scenario at 128 kbps. There's no need to ever use that codec at that bitrate. It leaves no room for error even if it would be transparent for a particular person. A 2014 Hydrogen Audio group test showed that the better codecs are challenging for most people at 96 kbps VBR, with Apple AAC leading the way as to what you could use practically. The people engaged in the testing were generally the cream of the crop compared to the general population. I don't think you could ABX 128 VBR Apple AAC on unexceptionally demanding music, but rather only in very special cases, if at all. Try it! :)

    Do the different rips from the same source, install the Foobar ABX Comparometer component, select the two versions of the tracks you are interested in, right click, select utilities from context menu, select the ABX comparometer, and you can do the ABX test. The software does the rest for you. And it tells you the percent chance you are just guessing when you are done. The default is 16 passes for a particular sample. It's really easy.

    The bravado, if any is to be derived from this hobby, is in my view not in how discerning your hearing is, but rather in your handle on the way things really are, and what you can do with that knowledge. Obviously there are people way, way past me in the Sound Science part of head-fi. In the rest of head-fi, after you do a little learning and experimenting, your jaw will drop when you read what people are saying. This is really a tight little discussion group among serious experts (with in some cases egos to match, I might add) with passersby like you and me dropping in every now and then. And if we say something too crazy we will draw flak.

    Last edited: Jun 10, 2018
  12. castleofargh Contributor
    ultimately, people need to test things for themselves. if audio insecurity is ruling over their lives, they probably shouldn't even bother with lossy codecs. because they won't enjoy music while paralyzed with the fear of "missing out" on something. one could say that lossy music will feel worst for them for real. for the wrong reasons, but still very real subjective impressions for them.
    for other people, if stuffing their DAP with more albums means making all the albums mono and limited to 13khz, they'll happily do it and feel that it's worth it.

    in a topic like this one, we go with the assumption that the user is seeking audible transparency for himself or at least differences that won't sound like horrible distortions. as not everybody has the same opinion on what is insufferable, or the same experience in identifying lossy artifacts, it just seems like the obvious thing to tell them to abx various formats and settings and decide what is the best sound/file size ratio for them specifically. some guy will notice a change at 128 and decide to go 1up in settings as a sort of security margin. personally I've been in situations where I could tell a difference, but didn't really have a preference, which is yet another can of worms when you start thinking about all the available codecs.
  13. bigshot
    He was interested in hearing how I determined the threshold where a codec becomes transparent. The way you do that is encode in CBR at way too low of a bitrate, then note the times where the biggest artifacts occur. Then you encode at progressively higher and higher bitrates using CBR and check those spots listening carefully until all traces of the artifact disappear. If you start at a higher bitrate, you don't know the areas of the track to focus on and you don't know what the artifact sounds like. It makes it much harder to find subtle artifacts. If you use VBR, the file redistributes unused bandwidth to prevent spot artifacting. hiding the codec's weaknesses. What I did was I determined the thresholds of the codec at CBR then used the transparent data rate with VBR to encode my entire music library. That way I know that the codec can reproduce my most difficult to encode music at CBR, and in case of an extreme situation, VBR could redistribute the bandwidth efficiently to ensure transparency.

    Does that make more sense? People don't seem to be grasping the logical process I was going though in my test.

    I'm Goldilocks. I'm looking for the balance of efficiency and quality. I want to know the point where a codec is as transparent as lossless and as efficient in size as it can be. "This setting is too big. This setting is too small. This setting is JUST RIGHT."
    Last edited: Jun 10, 2018
1 2 3 4 5 6

Share This Page