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

Setting Up An ABX Test + Simple Guide to Ripping, Tagging & Transcoding

First
 
Back
1
3 4
Next
 
Last
  1. reihead
    Great info, thanks!
     
  2. zilch0md
    You're a natural born teacher, Brooko. This is nicely done!
     
  3. Brooko Contributor
    Thanks guys - most of it's what I've learnt since I've been here.  And I really just wanted to have somewhere I could point people who are prepared to "enlighten themselves" - without having to keep typing the same instructions.  So now a simple link to this thread makes it a lot easier.
     
    Did you guys take the test - and what did you think?  Was it what you expected?  An eye-opener? (it was for me).
     
    What peeves me at the moment is that the few (very vocal) detractors of the sound science section simply refuse to actually test themselves.  They could effectively end the debate if they could prove their case.  I'm open to the idea that maybe someone can differentiate (consistently) high bitrate lossy from lossless - but to date, haven't sound a single person who can prove it (independently).
     
    Unfortunately the myth keeps being perpetuated ....
     
    a_recording likes this.
  4. zilch0md
    I'll not trouble myself to take the test, as I already believe I cannot hear any differences between the best lossy encoding and WAV, but I choose to consume additional storage space for WAV files (and that's no imposition for me) on the knowledge that absolutely no one disputes the abilities of the WAV format (until we get to the subject of bit depth and sample rate).

    For me, the choice to use WAV is just about peace of mind. I use battery power with all gear that allows it, and balanced cables with amps that allow it, not because I can honestly testify that I hear a difference, but again, just for the peace of mind that comes from "believing" its better.

    For sound quality, you just don't see many people arguing for the use of AC power converted to DC with a PSU as better than getting DC straight from a battery pack. And you don't see many people arguing that the sound quality of single-ended cables is inherently better than that with balanced cables. Similarly, fans of MP3 don't argue that it sounds better than WAV.

    :)

    But I'll say it again, I don't believe I can hear a difference between a "good" MP3 and a WAV file.

    Mike
     
    Brooko and landroni like this.
  5. Brooko Contributor
    And that I can completely understand Mike
     
    Cheers
     
    [​IMG]
     
  6. zilch0md
    Why can't everyone be normal, like us!    
     
    [​IMG]
     
    (Sarcasm!)
     
  7. FlyingKiwi
    Hi
     
    This is a great guide, thank you.  But - and probably it is my ignorance - there seems to be a bit potentially missing.
     
    When you play two different samples through Foobar with two different sample rates/depths, what does Windows do before feeding it through one's sound card and into one's headphones.  Sure, if a person has a DAC, this can all be avoided, but I'm trying to help a friend who has yet to be persuaded of the need to invest in an external DAC, etc.
     
    Does the Playback app thingy in Windows automatically convert all digital music to a standard frequency and bit depth?  If it does, how can this be over-ridden (assuming it even can!)?
     
    Many thanks for your further advice on this point - like you, I'm very keen to have people empowered sufficiently to be able to prove to themselves the validity (or otherwise) of the differences between different sample rates.
     
    Cheers
     
  8. Brooko Contributor
    Unfortunately I'm not as well versed with what the Windows mixer does - as I've always used WASAPI, and had the sample rates changing on the fly (bit perfect).
     
    However if you're doing a basic comparison - say any MP3 or AAC with the same bit-rate, and lossless redbook - then in reality both (depending on the set-up) would be showing 16 bit, 44.1.
     
    The other way of doing it - if I wanted to compare lossy with redbook with high res would be:
     
    • Use high res master (24/96) = file A
    • Dither copy to redbook (16/44.1), then redither back to 24/96 = file B
    • With original file A, transcode to aac256 (again using dither to set 16/44.1).  Then transcode back to 24/96 = file C 
     
    In effect you then have 3 x 24/96 files - but because of the dithering and transcoding applied, you can still abx - and check for audible differences.  If you can't discern any, then you know the lossy and redbook are essentially transparent with the original master.  The key though is to always use the same master as the initial start point.
     
    Again though - this is just my understanding.  there will be guys in the Sound Science section with far more expertise with this.  Might be a good idea to check with them as well. 
     
  9. FlyingKiwi
    Hi
     
    Thanks for this fast reply, and it might make sense to ask in the Sound Science section too.
     
    But, a quick follow up question to your comments.  How do you know, with the WASAPI interface, that the sample rates are being fed direct into the sound card's DAC without any change?
     
    I added the WASAPI component to Foobar, but if I now specify one of the WASAPI outputs, there is a requirement in its options to specify the output data format.  My sense is this will override the data format of the music clip, and I don't know how to ascertain exactly what is coming out of Foobar and into the sound card.
     
    I do understand that one could go through the processes you suggest as a work-around, but then I'd slightly worry that if a person was then detecting any difference in the files, it would be the result of a double resampling/transcoding and the filter artifacts caused by such a process!
     
    Many thanks for any further thoughts you might have.
     
  10. Brooko Contributor
    As I understand it, WASAPI is outputting a bitperfect stream, and is taking control of the audio output - bypassing the Windows mixer.
     
    At the same time though - my NFB-12 is an upsampling DAC/amp unit - so may well be upsampling everything on the fly to 24/96.  I'm not sure really whether it upsamples everything - or sticks to what it is fed.
     
    My suggestion again is to simply try it - without worrying about the bitrate.  If you can't tell a difference then it doesn't matter anyway.  If you can differentiate (which I doubt will happen) - then next step would be to look into it further. 
     
  11. FlyingKiwi
    Yes, I think that may be a suitable compromise.  If no difference is heard, then the matter is moot; in the astonishing scenario that a difference is detectable, one needs to research a bit further.
     
    Thanks for your further comments.
     
  12. byasini
    Nice tutorial.
    Thanks!

    But I think many of us wish to hear different between lossless and lossy while there aren't any different to our ears.
    For me as I am hearing mp3 since 90s I prefer to listen to a music with more details even if there is no more details than my previous mp3s.
    Now I wish I was 18 and I was at the maximum hearing age!

    Thanks again this was one of the best tutorial I read during past 3 years.
     
  13. arnyk

    ...surprise surprise, i cant tell the difference between flac and mp3 
     
    Great post, but a (probably pedantic) point of order.  
     
    The world is full of ABX tests with negative outcomes, and that just reinforces the idea  that nobody can hear $#!^ in ABX tests. However, that is often reality with many tests of really good stuff.
     
    I find it helpful to check the boundaries of what can and can't be done. For example we all know that there are low bitrate MP3s whose artifacts are easy to hear. Why not do a little experimentaion and find what that boundary is for you?
     
    Also there are often samples on the web that demonstrate that even high bitrate MP3s can be detected with the right music and your particular encoder.
     
    sonitus mirus likes this.
  14. Brooko Contributor
    Good point Arny.  I know with my own tests - for the majority of files 192aac was pretty much my limit - so I simply gave myself some headroom, and started encoding for my portable players at aac256.
     
  15. landroni
    To add to the discussion... In the lossy vs lossless debate, I find this take very instructive as to the nature of lossy compression (in 3 parts):
    https://passionforsound.wordpress.com/2011/07/30/understanding-mp3s-and-other-compressed-music/
     
     
    For Linux, Monty of Xiph recommends a command-line ABX utility called Squishyball:
    http://people.xiph.org/~xiphmont/demo/neil-young.html#toc_cbtpeadb
     
     
    However on Linux both Alsa and PulseAudio by default seem to be tampering with sampling rates (i.e. they resample on the fly), so it's an uphill battle to send bit-perfect data to the soundcard, probably a USB DAC. It's probably not a good idea to send 96 kHz data to your laptop soundcard, anyway.
     
    Any ideas how one would set up blind ABX testing on the X7? That would be a very interesting experiment to do, and I'm wondering if it's possible to leverage Android and some specific app/plugin towards this task...
     
First
 
Back
1
3 4
Next
 
Last

Share This Page