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

WAVs sound better than FLACs, which sound better than 320kbps MP3. 192Khz/24bits sounds better than 44Khz/16bits. TIDAL Hi-Fi sounds better than Spotify Premium. Tubes sound better than Transistors.

Discussion in 'Sound Science' started by deafmutelame, Apr 26, 2018.
2 3 4 5
  1. deafmutelame

    ...ok, 320Kbps MP3 might not pass an ABX test with certain 'killer' samples but...now that I got your attention... :dt880smile:

    The funny thing is that these myths are SO prevalent among casual listeners (as I was once too, aeons ago...)

    I think something should be done about this and what a better place that a science-saavy forum like this one.

    So I propose that we begin to gather links to papers and posts that explain the truth behind these myths, so that we can point laymen to this thread and help them to dispell these assertions.

    I will be updating this first post with links to ABX tests, articles or scientific papers that you find useful.

    I'll start with a nice one, straight from bigshot's signature:

    24/192 Music Downloads ...and why they make no sense

    LAME (MP3 encoder)

    List of recommended LAME settings


    ABX High Fidelity Test List: Is your audio system really ready for lossless sound?


    ABX Tests Prove Hi-Res Audio Is Legit!


    Conclusive "Proof" that higher resolution audio sounds different
    Last edited: Apr 26, 2018
    skwoodwiva likes this.
  2. ev13wt
    Some people prefer lower than posibbly fidelity. Casual listeners like too much bass and "colored" sound.Consumers that would like to understand more, and not be bent over with a usb and 3.5mm hanging out, will find your post. But we will not "correct" an entire industry, within capitalism, with any posts.

    Some people want to be lied to, some need a 5USD DAC in a 10000 case...
    Some actually like tubes - as a hobby - as a technology - as a way to say: "Yes, that is an aduiophile amplifier, pretty cool, huh?"

    I like tape.
  3. deafmutelame
    So true. There's some people whom would genuinely like to understand more. To those I'd like to address this thread.

    This is the kind of post that I would have liked to find years ago, when I was wondering how and why to rip to MP3 my entire CD collection, for instance. When disk space was costly and scarcer than today you had to plan those things ahead...

    By that time it used to be said that LAME --alt-preset standard was considered transparent enough. I wasn't sure of what to do until I found this thread from 2001 at hydrogenaudio:

    List of recommended LAME settings


    Another resource:

    ABX High Fidelity Test List: Is your audio system really ready for lossless sound?
  4. deafmutelame
  5. bigshot
    A gamelan gong would be better to do that with than keys jingling. But I guess no one has a gamelan gong. I think I have a CD that has some gamelan music on it. Maybe it's the one album in my collection that would benefit from high sampling rates.

    Also, Amir has admitted that in a test, he chopped up short bits so he could focus on one particular bit of the sound. He didn't play it through the track the way a person generally listens to music. I listen to music with my system. I don't zoom in on bits. I have no doubt that someone can game the test to win. You can pump a square wave through just about any piece of consumer equipment and get some nice audible distortion. But a square wave isn't music. I would like to see results like this showing that the difference can be heard in normal music listening conditions. I don't believe it can. Normal music listening conditions are my benchmark. I don't need to have perfection under circumstances as specialized as being on the moon.
    Last edited: Apr 26, 2018
    snellemin likes this.
  6. amirm
    "Normal music listening conditions" do not include any comparative testing. So by your definition no such test can exist.

    Please, learn something about the topic before making these comments. There are plenty of papers, international standards, etc. on proper listening test procedures. You seem to think the only good test is one that generates negative outcomes to match your narrative.
  7. amirm
    They HA guys often work backwards to make the data match their point of view. Take this example chart that used to be in their Wiki:


    As you see, the quality graph never achieves transparency at score of 10. I got into an argument with one of their members saying it did achieve transparency. I showed him the above graph. He went back to the people there. Their answer? Let's delete the graph! And that is what they did. Never mind that the graph was correct. There just isn't a lossy codec which is transparent for all content and all people.
  8. bigshot
    I'd be more than happy to give you some parameters so you can determine whether something performs well under normal listening conditions if you'd like.
  9. amirm
    Go ahead. Or are you waiting for a red carpet to be rolled out first?
  10. bigshot
    Just wanted to know you were receptive.

    Normal loud listening volume would be below 80dB- it might even be a good idea to do the tests at a normal volume level too- say 50dB. Volume levels should be matched and fixed. The listening test should be conducted with a variety of different kinds of music- classical, chamber, jazz combo, choral/vocal, rock, electronic. The two sources should be playing simultaneously, direct A/B switchable. Subject can choose when to switch. Music should play through- no stopping, looping, skipping, scanning or scrubbing. The position of the sources on the switch should be randomly assigned blind with each sample played. Multiple test subjects with multiple tests performed for each subject. Subjects should identify the channel on the switch that has the best sound quality to them. Record the choices compare to the sources and see if there is a pattern when all the responses are averaged.

    I'm not experienced in testing procedures, just sound, so feel free to add things that would offer further refinement!
    Glmoneydawg likes this.
  11. ev13wt

    Indeed, a great way to have all sides of the story available to people who seek information.

    Without going into the debate about high res, there is no denying something was heard in the key tests. With all info available, an informed descision can be made.

    Do I need a record player? Yes.
    Do you need a DAC? Of course.
    High Res? Sure!
    Tubes? Yep.
    Big ass reel to reel 2 Track? Well, duuuhhhh.
    Giant speakers? Well... naturally!
    Low-fi cassette tape demos? Hell yea!

    It really easy once you understand how it works. Unlike before 1990, when most things could be prodded at, picked apart and one could see what lever pushed what - or things of that nature.
    Last edited: Apr 27, 2018
  12. bigshot
    The easiest way to determine whether a file is high sampling rate or not is to open it up in a sound editing program and look at the frequencies above 20kHz.
  13. amirm
    Come again? How is that "normal music listening conditions?" When you are listening to music are you in the frame of mind of comparing it to an imaginary alternative?

    Let's say there is a difference in A vs B. You listen to them and hear the difference in such a blind test. Is that how you vote? I bet not. Half the time you will second guess yourself wonder if the difference is real versus imaginary. This doesn't happen at all when listening without evaluating.

    Since these concepts are foreign to you, think of taking a class at college versus taking an exam. One is a lean back experience. The other, not.

    Our goal in creating such tests is to overcome the barriers created by them. This is why we give tools to people to select clips. This is why we let them change volume. We want to compensate for negative aspects to get to the truth. That is, if there is a difference, we want to know it. We don't want to do everything in our power to get a negative outcome.

    Here is the output of the most popular ABX tool, the plug-in to foobar2000:


    You see the "set start" and "set end" for creating a clip?

    Do you see the Position progress bar so that we can find the specific section?

    Do you understand the notion of dynamic distortion that appears on say, high frequency transients and not elsewhere?

    So no, our goal is not to stick our head in the sand and create test conditions that are less revealing. We want to do our best to find audible difference. Because once there, we can fix the issue. We can buy gear that is transparent. We can get content that is transparent. And none of this has to cost much.
    snellemin and skwoodwiva like this.
  14. Glmoneydawg
    This is excruciating,but it looks like you 2 maybe co-operating.....looks like a great idea though...make it happen:)
  15. bigshot
    I think it would be a good idea too. Listening to music is what we do. Everything should be the same as that, except for switching between samples. If it's hard to tell the difference doing that, it isn't worth worrying about.

    I don't see any point constructing artificially convoluted situations designed to break things. I don't care if it doesn't properly reproduce frequencies higher than a bat can hear, square waves or tiny fractions of time that don't exist in recorded music. I just want a system that plays back music perfectly.
    Last edited: Apr 26, 2018
    Glmoneydawg likes this.
2 3 4 5

Share This Page