Why were CDs recorded in 16-bit/44.1khz?

Discussion in 'Audio Hardware' started by MZ_RH1, Feb 5, 2017.

  1. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    I am pretty good/fast at research -- and just read through many documents on the audiblity of polarity (just 0 and 180deg). And after doing the reading, I have to say that there are too many variables even in the case of someone APPARENTLY detecting the 'phase' (polarity), that it can be some kind of external effect. I am not reaching at straws, but rather looking at every part of the signal chain, and there are just TOO MANY chances for a different phase to produce different effects. Most devices are not totally linear (even though electronics circuts have gotten better until about the last 10-20yrs, where they can be near perfection), but even the headphones/speakers can act differently on asymmetrical waveforms. Even amplifiers typically have push pull devices which aren't totally symmetrical -- and the distortion effects (mostly intermod at higher freqs) differences can be audible. However, at 1000Hz with a sine wave (or even a mixed 1000Hz/2000Hz intermod) for a really good amplifier can really be inaudible, but we are talking about the general case where there is more in the chain than just a near-perfect amplifier.

    Of course, perhaps the most extreme case is that of AM modulation, where even 200% overmodulation in the positive direction doesn't necessarily clip, while 100% negative modulation is the limit because transmitters cannot produce negative power (suck power in the sense of a modulated signal from the environment.) In the case of negative overmodulation, there is clipping. So, an AM transmitter will produce different distortion from one polarity to the next for overmodulated signals. Similar kinds of differences can be true for diaphragms, except to a lesser degree and without such a sharp cutoff.

    Even good headphones can produce different kinds of distortion on an asymmetrical waveform between 0 and 180deg phase shift.

    The only case where it might be possible to prove detection of phase reversal might be for low frequency transients, but even that will likely not be detectable for sine waves.

    Also, the only non-LF test that I could see that is truly valid would be a sine wave test -- but the idea of phase reversal in that case is silly, because the phase has to be measured against an absolute time, and our hearing just does not work that way.

    So, from a true scientific/engineering standpoint, I am not able to fully discount the possibility of truly detecting polarity with the human hearing system, but there are so many caveats to even a statistically valid result of detection, that it is almost impossible to make the claim.

    I do have an idea for an experiment -- given that someone is really interested. In this experiment, all of the data needs to be correlated, and any one test cannot be defininitive. The scenario is like this, and even this is purely statistical:
    -----------------------------------------------------------------------------------------------------------------------------------

    Run multple passes on DIFFERENT different equipment combinations, and each pass runs multiple pieces of music in multiple genres along with tests for synthetic signals.

    If the tests between different equipment tend to correlate (that is, the test using different individuals, but similar on certain combinations of equipment), then the results tend to be more dependent on the equipment characteristics.
    If the tests between different equipment tend to NOT correlate(that is, the test between different individuals doesn't seem to show consistency between different combinations of equipment), then the results tend to be less dependent on the equipment characteristics.

    Correlation results about specific equipment compnents, for example sets of headphones and any other transducer in the chain (e.g. DACs) need to be included -- just in case there is some dependency on equipment characteristics Then, the enabler of the ability to detect might be narrowed down.

    If the abillity to detect appears to be random based upon the equipment (but there appears an ability to detect by the statistics), then it is more likely that the determining factor is human perception, but even then it isn't proved -- just more likely.

    If the ability to detect is random based upon the equipment, but also that it appears that the 'detection' is not happening, then there is probably not any ability to detect polarity.

    (This is not a complete description, but the idea that the signal, signal type, equipment and individuals (maybe other things) all have to have a large enough sample set to get statistically useful results.)

    This would be a monsterous project, and would resolve a matter that most people just don't care enough about given the cost/complexity -- but SOME students might want to do so for a research paper?

    The most expensive parts of the project would be 1) enough indidviduals to participate, 2) enough variations in equipment. Of course, the numerous signal sources and types wouldn't be a cake walk.

    MOST IMPORTANTLY -- those running the experiment must not be too biased, perhaps even add some aspect of being double blind to further avoid bias.

    The project wouldn't either be fun, and wouldn't resolve a matter all that important to me - but maybe some researcher could put together a much more formal version of what I am trying to describe!!!

    John
     
  2. Holy Diver

    Holy Diver Senior Member

    Location:
    USA
    Just remember, all you need is a CD and a player. That's all that matters. :)
     
    enfield likes this.
  3. Thomas_A

    Thomas_A Forum Resident

    Location:
    Uppsala, Sweden
    Regarding a study of polarity, I think there is no one willing to conduct that since it is probably made already.
    Cited from a paper 1982: "It is known that the inner ear possesses nonlinearity (akin to an acoustic half-wave rectifier) in its mechanical-to-electrical transduction, and this would be expected to modify the signal on the acoustic nerve in a manner which depends upon the acoustic signal waveform, and so upon the relative phase relationships of the frequency components of this signal. Some of these effects have been known for over 30 years, and are quite audible on even very simple signals."
     
  4. Steve Hoffman

    Steve Hoffman Your host Your Host

    People, let's move on from absolute polarity, can we? Or start another thread on the subject.

    Those of you who are blissfully unaware of it, good, you're lucky. Those of you with Polarity Buttons on your CD player remote, DON'T PUSH THE BUTTON TO FLIP IT!!!

    Oops, too late.

    Now you too are aware of the giant difference it can make in your music and will never sleep soundly again until you make a note as to which polarity position sounds best and stick it on to every CD in your collection.
     
  5. Thomas_A

    Thomas_A Forum Resident

    Location:
    Uppsala, Sweden
    I 'll leave it.

    The purpose was just to demonstrate that yes, it is audible during certain conditions, but extremely rare if ever when playing music as we do usually do it.
     
  6. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    Don't you need headphones and an amplifier of some kind too? I'm not good at humor, and I apologize for the misconstruing otherwise :).

    There are so many papers in both directions, and papers that explain away other papers, that is why I suggest doing a fully controlled experment where all of the various possibillites are tested for probability. I don't even believe that they had transducers in 1982 (or even today) that are good enough to prove a hypothesis without doing a pretty large statistical analysis. Conjecture is good enough to get started, and if someone really wants to demonstrate (not really prove, but be able to get a consensus) the likelihood, then the fully outfitted experiment is the way to go.

    It isn't good enough to say that there is a liklihood that hearing can detect polarity unless the other variables are removed. Even if there are some stats that show it with really good headphones, there are really too many variables without testing other kinds of headphones which might have different kinds of nonlinearity.

    I remember a kid who was talking about during a hearing test that they were bragging about having so sensitive hearing that they could hear 'nothing.' I also took the test, and found that I could hear the hiss (the tone was zero) also. He was claiming to be able to hear the tone, when actually he was likely hearing the hiss in the apparatus. I don't trust very uncontrolled tests -- and I am only trying to help either side learn about their beliefs on this matter.

    John
     
  7. Steve Hoffman

    Steve Hoffman Your host Your Host

    What did I just say, kids?
     
    TimB and dalem5467 like this.
  8. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    I apologize -- I didn't see your post. I would have complied.
     
  9. Steve Hoffman

    Steve Hoffman Your host Your Host

    I'm not truly serious. The subject deserves its own thread, that's all.
     
  10. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    I wanted to follow-up about this matter of using the right version of the sox attenuation command. I have this recording in my collection that has been a thorn in my side for a LONG time. I know that it was DolbyA encoded, but I couldn't get rid of the 'splats' in the signal (they weren't as bad as a really bad inversion, but sounded ugly.) Over and over again I had modified my decoder with all kinds of attempts to 'scope probe' in software to figure out it.

    Just a few hours ago, a 1yr quandry has been solved.... It was some mild clipping in the conversion from the flac file in sox. I had been using 'gain -6' to mitigate any chance of clipping, but the troubles had already happened. It is apparently important (to mitigate clipping on input) to use the -v 0.5 switch instead of the 'gain -6' command later on. I did a quick scan of the sox source, and it appears (could be wrong) that sox internals in the '24 bit signed integer world' instead of 'floating point world'. I have wasted a lot of time in trying to figure out what was wrong with my decoder in that regard, and NOW I have a perfect rendition of Super Trouper!!! (Who cares about 'Super Trouper' per se? Not me -- just that I have figured out the problem.) My decoder just did some moderately stupid things when trying to decode -- it wasn't a severe algorithm problem that caused that specific trouble... (There have certainly been severe problems in the decoder, but that wasn't caused by one of those.)

    So, the moral of the story -- try to use '-v 0.5' if you need attenuation on the input, and not the 'after the fact' 'gain -6.0'.

    John
     
  11. Hannah

    Hannah Only love can conquer

    You may want to read Daniel J. Levitin's This Is Your Brain on Music: The Science of a Human Obsession. A great book full of surprising brain/music science.

    The book has been discussed here in the past, but it kind of belongs in this thread.
     
  12. AvineshChitrakar

    AvineshChitrakar New Member

    Location:
    kathmandu, nepal
    Most answers were broad, the following are answers given in the marking scheme if CIE A Level exams provided by Cambridge University professors.

    • resolution is the number of distinct values available to encode/represent each sample
    • specified by the number of bits used to store/record each sample
    • sometimes referred to as bit depth
    • the higher the sampling resolution, the smaller the quantization error
    • a higher sampling resolution results in less distortion of the sound
    • usually 8 bit, 16 bit, 24 bit or 32 bit
     
  13. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    On the sampling resolution vs. distortion and sampling resolution vs. quantization error... Dither and appropriate filtering resolves those matters -- so even though there is a minimal difference in audio distortion, it is generally nil. The normally detectible difference results wrt sampling resolution is similar to hiss-type noise.

    An example of dynamic range being incredibly bigger than implied by the sampling resolution would be a cell system receiver. They have to be able to work between low microvolts up to big fraction of a volt, and still stay linear WITHOUT significant AGC. What kind of ADC do they use? 12 or 14bits (maybe 16bits -- not sure.) They get approx 100k:1 dynamic range with 16k:1 resolution in the ADC!!! You can get amazing linearity with low bit resolution... How do you do it? dither & appropriate filtering. Dithering&appropriate filtering (sort of) spreads the errors and turns them into noise (in a way.) With certain techniques, they can process the signals way below the total noise level. If the cell receivers werent somewhat linear, there would be distortion products all over the place, and nothing would be usable.

    Audio is NO different in this regard. If you use a 12bit A/D correctly on audio and compare it with 16bit results, all else being equal and done correctly, the big difference is the hiss. (again, because the dither plus special filter, 'kind of' converts the 'stairsteps' into 'hiss'.)

    Done correctly, 'stairsteps' are effectively null & void.

    John
     
    sunspot42 likes this.
  14. Scott Sheagren

    Scott Sheagren I’m a Metal,Rock,Jazz Fusion,Gaga type of guy.

    Location:
    06790
    it think cds there self were not the problem.it was the players at the time.players and dacs now sound amazing.
     
    apesfan, The Beave and enfield like this.
  15. vwestlife

    vwestlife Forum Resident

    Location:
    New Jersey, USA
    Au contraire:

    [​IMG]
    (from Stereo Review, 1997)
     
  16. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    I'd suspect 90% of the problem with 'screechiness' was bad or non-existent mastering... Period...
    Noise reduced tapes without NR decoding DO NOT SOUND GOOD.

    John
     
    ted209 and Scott Sheagren like this.
  17. Joint Attention

    Joint Attention Forum Resident

    Location:
    Gig Harbor, WA
    I don't recall where I read this, but supposedly the 74 minutes was based on the Sony chairman's wife's favorite performance of Beethoven's 9th, which is this one:

    BEETHOVEN Symphony 9 Naxos 8.111060 [SH]: Classical CD Reviews - March 2007 MusicWeb-International

    Karajan was a consultant on the development of the CD, but his recordings of the same symphony clock in at under 74 minutes.
     
    apesfan likes this.
  18. Black Elk

    Black Elk Music Lover

    Location:
    Bay Area, U.S.A.
    No, this has been discussed earlier in this very thread:

    Why were CDs recorded in 16-bit/44.1khz?
     
  19. vwestlife

    vwestlife Forum Resident

    Location:
    New Jersey, USA
    DAK had that all figured out in 1985:

    [​IMG]

    (from Stereo Review, September 1985, page 67)
     
  20. John Dyson

    John Dyson Forum Resident

    Location:
    Fishers, Indiana
    QUOTE: <Stereo review article>

    Sorry -- I hated to quote the entire thing -- but I havent't found all that much of a dropoff between 5-10kHz on GOOD old vinyl when compared between DolbyA semi-original vs. GOOD old vinyl that I have ripped copies.
    If you do consider that DolbyA has an effective boost starting at about 3kHz (it isn't really a boost as DolbyA is essentially flat at 0dB), but it is an evil-fast multi-band compression from lower levels, then there would be a difference between undecoded material and properly handled vinyl creation. Decoding-on-the-cheap by doing a bit of EQ (-3 or -6dB at 3kHz and/or -3dB at 9kHz), still leaves a characteristic sound quality (at times, depending on material -- difficult even for me to detect, but other times very noticeable) & sometimes an UGLY HF boost. HF intensive pop would be a bit more difficult to detect (other than greater HF ambiance) than more mellow material with SOME HF.

    Back then, I'd suspect that they were clueless about the lack-of-decoding.
    There shouldn't be very much difference between a properly mastered CD and properly made vinyl (vinyl often has a bass cut and sometimes a bit of HF compression or EQ when needed.) The difference would be noticeable, but not terribly significant.

    Old CDs were often harsh, but the reason wasn't 100% vinyl rolloff.

    John
     
  21. enfield

    enfield Forum Resident

    Location:
    Essex UK
    Lets not forget that early CD's were sometimes mastered with an extra bit of HF. As if to emphasis the clarity and sharpness of the format over traditional formats.This soon stopped.Unfortunately compression and brickwalling took its place.
     
    sunspot42 likes this.
  22. vwestlife

    vwestlife Forum Resident

    Location:
    New Jersey, USA
    Are you sure you're not confusing that with older CDs that use pre-emphasis, which modern CD players (and CD ripping programs) may or may not properly decode?

    Pre-emphasis - Hydrogenaudio Knowledgebase
     
  23. Dennis Metz

    Dennis Metz Born In A Motor City south of Detroit

    Location:
    Fonthill, Ontario
    True:cheers:
     
    ted209 and Scott Sheagren like this.
  24. Scott Sheagren

    Scott Sheagren I’m a Metal,Rock,Jazz Fusion,Gaga type of guy.

    Location:
    06790
    wow no kidding.so why does vinyl have that thicker lows and mids and brassy cymbals that sounds better?
    it took along time for my cds to sound just as good if not better then my vinyl now.thanks to my amazing dac.
     
    The Beave likes this.
  25. vwestlife

    vwestlife Forum Resident

    Location:
    New Jersey, USA
    Distortion. To many people's ears, a little bit of distortion adds "warmth" at lower frequencies and "shimmer" at high frequencies.

    If you've grown up listening to music with that kind of distortion in it, then suddenly removing it, such as with the switch from LPs to CDs, can make the audio sound "cold" and "thin", even though it is actually more accurate to the original recording.

    Similarly, when color TV first became popular, many older people complained that it "hurt their eyes". TV manufacturers responded by continuing to offer black & white console TVs as a "Senior Citizen Special" well into the 1970s.
     

Share This Page

molar-endocrine