192kHz/24bit vs. 96kHz/24bit "debate"- Interesting revelation

Discussion in 'Audio Hardware' started by mindblanking, May 10, 2013.

Thread Status:
Not open for further replies.
  1. Metralla

    Metralla Joined Jan 13, 2002

    Location:
    San Jose, CA
    I do agree that (for me) it's often hard to discern the differences between very good DACs.
     
    Spek likes this.
  2. TommyTunes

    TommyTunes Senior Member

    I say that about most gear at the same price point. It more about flavors than quality. Let's face it, there is not going to be a big difference if you were to compare three $10K pre-amps, they would all be good, just different. Same with DAC's, Cartridges etc.
     
  3. Metralla

    Metralla Joined Jan 13, 2002

    Location:
    San Jose, CA
    You are so right. But DACs do seem to be particularly vexing.
     
  4. Spek

    Spek Well-Known Member

    Location:
    DFW, TX
    I'm not sure why different "flavors" would be necessary or even desirable when it comes to DACs. Wouldn't the only qualification for a DAC be that it's audibly transparent? Once that threshold is reached, by definition, what further "improvement" can be made?
     
  5. Metralla

    Metralla Joined Jan 13, 2002

    Location:
    San Jose, CA
    The analogue stage gives a flavour
     
    morinix likes this.
  6. TommyTunes

    TommyTunes Senior Member

    DAC's have an analog stage and just like every other piece of gear the analog stage colors the presentation. Therefore some may be darker sounding or be brighter or offer a more recessed or forward sound etc.
    Please decribe "audibly transparent" because in all my years everything has a unique sound. I've never heard any two items that sound exactly the same, at any price point.
     
  7. Spek

    Spek Well-Known Member

    Location:
    DFW, TX
    By "audibly transparent" I mean a DAC that the noise and distortion added by the DAC are imperceptible to the listener. In other words, a DAC that faithfully converts the signal without adding anything to it. As far as I know this is not uncommon, and it doesn't cost all that much to build or purchase. That's personally all I'm looking for in a DAC.
     
  8. Jim T

    Jim T Forum Resident

    Location:
    Mars
    ----------------------
    Shure and Rode have recorded samples of their mics on their websites where you can hear the differences, and ribbons are remarkably warm. They are not necessarily accurate, but they sound great. I just wish they were more affordable.
     
  9. Jim T

    Jim T Forum Resident

    Location:
    Mars
    --------------------------
    The Bifrost dac has me really interested.
     
  10. Danglerb

    Danglerb Forum Resident

    Location:
    Orange, CA, USA
    Audibly transparent, that condition when two pieces of gear sound unique, yet the listener is unable to identify which is which without non audible cues better than random chance.
     
  11. Music Geek

    Music Geek Confusion will be my epitaph

    Location:
    Italy
    Oh yes.... I can do even better and I can even tell the difference in sound between two instances of the same kit... but only when one of the two is broken :)
     
  12. Hobbit13

    Hobbit13 New Member

    Location:
    Netherlands
    In this lengthy discussion, has anyone performed an unbiased comparison?
    • Take an audio system that is fully capable of reproducing 192KHz, including speakers that can reproduce up to 100KHz (those are quite rare!!)
    • Take a 192KHz recording, confirm via spectrum analysis that it has frequencies up to nyquist
    • Resample the recording to 96 and 48khz via SoX, and upsample the result to 192KHz (upsampling is important, to rule out difference in dac performance for different sample frequencies!)
    • Use blind testing software to see if you can identify which recording was resampled, and which was native 192KHz
    I once did such a test using Foobar ABX Comparator, and could not identify between 48 and 96, but my speakers do not go up to 40KHz. The 96Khz resample sould have some ringing artifacts around 46Khz, but those are so much more quiet than the music in the audible range, that I'm pretty sure no one can identify them, even if they would hear up to 40KHz. For 48KHz resample the ringing would be closer to the audible range, but still low in amplitude.
     
    Spek likes this.
  13. LeeS

    LeeS Music Fan

    Location:
    Atlanta

    You don't need to reproduce ultrasonics to get the benefits of hirez playback.
     
    jmathers likes this.
  14. DragonQ

    DragonQ Forum Resident

    Location:
    The Moon
    I plan to do this with my new audio setup (won't have time for a few weeks though as I'm moving). My speakers only go up to ~25 kHz but, as LeeS says, the benefits of hi-res audio are often touted as affecting the 20 Hz to 20 kHz region, despite the fact that any audio signal in that range should be able to be represented exactly by sampling at 40+ kHz. So all you need are a DAC capable of 96+ kHz playback, some 96+ kHz audio material, and Foobar.

    I highly doubt I'll be able to tell the difference but, as a scientist, I intend to try my darnedest. Of course, if I can't tell the difference, then that just proves that I can't do it with my setup, nothing more. That's the problem with negative results, and is why there is such a thing as a "burden of proof" and the "null hypothesis".
     
    Spek likes this.
  15. LeeS

    LeeS Music Fan

    Location:
    Atlanta
    Unless you understand what Kuncher is saying and then 192khz is required.
     
    jmathers likes this.
  16. DragonQ

    DragonQ Forum Resident

    Location:
    The Moon
    I was speaking more generally but, sure, 192 kHz is do-able.

    Are you suggesting that 96 kHz is no better than 44.1/48 kHz then, and that only 192 kHz is any better?
     
  17. LeeS

    LeeS Music Fan

    Location:
    Atlanta
    I'm saying that based on the evidence presented, you need 192khz. 44.1 or 48khz is clearly inadequate. 96khz is quite good.
     
  18. DragonQ

    DragonQ Forum Resident

    Location:
    The Moon
    I see. Have you read the Hydrogen Audio thread discussing these papers? I was going to check what other scientists made of the conclusions but one paper hasn't ever been cited (in 5 years) and the other doesn't even exist on Web of Knowledge. :sigh:
     
  19. Spek

    Spek Well-Known Member

    Location:
    DFW, TX
    There's no doubt that Kunchur's tests are interesting in showing that humans can resolve inter-channel timing differences of roughly half of what was thought before (as I understand it). Where I fail to make the connection is why properly-dithered 16/44.1 wouldn't be able to resolve these timing differences. As a couple of the guys on Hydrogen Audio said, it may have more implications on jitter levels than on "low-res" vs. hi-res. As I posted before, it seems like assumptions were made in Kunchur's section on "Implications for Sound Reproduction":
    It appears his summary of how it is "known that the bandwidth requirement for sonically transparent audio reproduction is higher than the 20 kHz" is based on Stuarts paper, which, as far as I can tell, is just because Stuart says so. I think more testing is needed, both to replicate Kurchur's results and to test other things besides a 7 kHz square wave.
    It's good that you're trying. I have tried many times but I also have a bias toward the null hypothesis, since that's what digital audio theory says should happen. It's true that quite often audiophiles make claims regarding hi-res that have nothing to do with frequencies outside the hearing range, but they also often say that ultrasonic frequencies have an affect on the audible band, so I don't see how this could take place unless your equipment can accurately reproduce these frequencies. As I understand it, this "mixing" of the frequencies is supposed to happen after the sound leaves the speaker (I may be wrong about that though).

    What's interesting is the negative effects these frequencies can have when we attempt to reproduce them. As was pointed out on the xiph.org article on 24/192, the ultrasonics in hi-res can be a liability during playback depending on the equipment. There is an interesting graph on that site showing typical IM distortion when a 30 kHz tone is played on an amplifier model with THD of about .09%:

    [​IMG]

    As you can see, although the 30 kHz tone is certainly not audible, the IM distortion reaches well in the audible range. There have been studies which tried to reproduce Oohashi's results but they have not been successful. Because of the effect illustrated above, it is suspected that this may have played a role in subjects being able to distinguish full-bandwidth from bandwidth-limited material. The effect is very slight, but there's no question it could be audible. I find this particularly fascinating, and really could theoretically make something sound "better." Just one theory.
     
  20. soundQman

    soundQman Senior Member

    Location:
    Arlington, VA, USA
    It's interesting that the OP chose to use a Meridian device (The Explorer) for his comparisons. Because Bob Stuart, one of Meridian's co-founders (the engineering/technical design partner) is on record as believing that 192/24 offers no sonic advantage over 96/24. Consequently, their digital systems down- or up-sample all signals to 96/24 for processing on the basis of this opinion. They made one concession by agreeing to handle 192/24 without downsampling only through their USB inputs. They have done this as a marketing concession to computer audiophiles who wished for this feature, not because they believed in its superiority. At least that's the explanation I got from my Meridian dealer.

    And indeed, I must admit that I don't hear a difference between the two rates on the same program material, but perhaps I didn't control all possible variables and perhaps my hearing acuity isn't capable enough.
     
    Spek likes this.
  21. Stephen Murphy

    Stephen Murphy Forum Resident

    Location:
    Edmonton Alberta
    You make an excellent point about the Meridian downsampling and upsampling automatically by its very design. The Benchmark DAC1 is another common unit that resamples everything to 110kHz. A person trying to compare files having different sampling rates like 96 and 192 khz would hear no difference with one of these DACs because both files would be played back at the same rate.
     
  22. Don Hills

    Don Hills Forum Resident

    Mmmm, popcorn...
     
  23. Don Hills

    Don Hills Forum Resident

    It's evidence that higher sample rates have less energy smear caused by the filters. It is not evidence that there is a sonic advantage to having less "energy smear". It would be easy enough to test: Take 2 copies of a 24/192 source. To one, add the amount of "energy smear" that would occur with a sample rate of 44.1 or 48 KHz. ABX the two copies.
     
    GetHappy!! and Spek like this.
  24. kevintomb

    kevintomb Forum Resident


    Not adding noise or distortion that are audible to a listener, nor altering its tonality or overall frequency response, when the listener is not aware of the unit, it neither adds nor detracts from the sound presentation.

    The way I read some of the descriptions here, sounds to me at least like we are describing differences in carts and speakers. Or at least used cars.....:D

    Agreed, everything has a very slightly coloration, although ive seen many say, removing the visual cues, or knowing which is playing often takes away a lof of the variations being described. For whatever thats worth.

    I tend to see Speakers, The room, then the listener to usually be the 3 biggest variables.
     
    drSeehas likes this.
  25. LeeS

    LeeS Music Fan

    Location:
    Atlanta
    It doesn't matter whether there are theoretical problems if the implementation is imperfect due to filtering issues. That still means 24/192 or 24/96 sound better than 16/44.1. I'm only concerned with what the final product sounds like. And all else being equal 24/192 sounds better than 24/96 which sounds better than 16/44. And I've done these tests over and over with live music recordings so I have the live source to compare with and no mastering at all. Just playback that is level-matched.
     
Thread Status:
Not open for further replies.

Share This Page

molar-endocrine