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

Thoughts on a bunch of DACs (and why delta-sigma kinda sucks, just to get you to think about stuff)

Discussion in 'Dedicated Source Components' started by purrin, Dec 5, 2013.
371 372 373 374 375 376 377 378 379 380
382 383 384 385 386 387 388 389 390 391
  1. goobicii
    but then you are hearing altered signal not original one
  2. Sorrodje
  3. azteca x
    44.1/2=22.05. Add in some wiggle room for filtering and whatnot and you get up to about 20kHz (20,000Hz) pretty reliably. Of course, I'd love to see a test where you demonstrate that you can hear differences ±0.1dB at 20kHz.
    The highest sample rate I've ever seen (for PCM) is 384kHz, and it's really just hardware support. There are like two things recorded in that format. No recorder can use the extra high frequencies. And as Lavry pointed out in that whitepaper, there is a time where higher sample rates introduce more distortion.
    So, how did you do this 1536kHz test?
    Also, lossy refers to whether or not a format preserves the original RECORDED signal. So if you can take the original WAV and then take a FLAC you bought off of Bandcamp, invert phase on one of them and get a complete null (silence), you have a lossless format. I don't think anyone would claim that there is a lossless recording method. Recording is a game of compromises and always will be.
  4. prot

    many called that filter "special sauce" or something like that and IIRC you did it too ... so I really dont understand what's wrong with someone asking just how 'special' that is ... or why is everyone jumping so hard on that guy for just asking some questions

    Anyway, got another burrito Q: guess it is safe to assume that the original filter does still exist as Theta's property .. are there any other current devices using it ? (theta or licensees, or...)
  5. azteca x
    Keith, if it helps: http://www.head-fi.org/t/764787/yggdrasil-technical-measurements
  6. goobicii
    with oversampling,a process that exist only becose the truth is that 44.1khz is too low samplerate
    what who seen,what records exist or what are current technology limitations is not my point,the point is anything less than 1536khz CANNOT play a sinewave with constant volume in human hearing range 20-20.000 khz.                personaly I hate the Lavry guy,he thinks he is some guru but he is just bad influence becose people who doesnt think and experiment themself follow him blindly,high sample rates are future,Lavry and his BS is just slowing the technological advancement
    simple,just create sinewave at whatever you consider highest freqency that should be reproduced accurately,lets say 20.000hz,create that sinewave at various samplerates and tell me at what point it becomes perfectly constant in volume without any loudness jumping
  7. AustinValentine
     @prot - This is the answer to why people are losing their patience with Diamondears. He's essentially been posting the exact same thing in this thread for over a month now without engaging any of the substantial replies that he's received. See inside the spoiler to see what I mean, with posts presented in reverse chronological order [with all of the insulting/bickering/trolling comments in-between removed]: 











    wahsmoh and prot like this.
  8. wmedrz
    It sucks that Purrin is MOT. I really enjoyed reading his opinion about limp d*ck craptastic gears :frowning2:
    HemiSam likes this.
  9. KeithEmo
    I've seen a few - very few - commercial 384k files (I even have two of them), but they are VERY few and far between. (I believe I found two websites that each had a few examples of 384k content.)
    I'm afraid I've got to agree that the white paper about "higher sample rates sometimes causing distortion" is really rather specious. (It's kind of like saying "a TV picture that's too good might make a poor quality projector actually look worse because it struggles to reproduce the unnecessary extra quality".) I cannot fault a content FORMAT for allowing it to reproduce the content more accurately than necessary. If the content producer is worried about 50 kHz noise affecting your playback system, then THEY should limit the bandwidth of their recording - it is not a problem for the recording to have frequency response that's simply better than necessary - unless you're counting on its limitations to protect you from poor production control. (And, if you know that your stereo has problems with a little 25 kHz noise, then it's up to you - or the manufacturer - to "protect it".)
    A sine wave that is converted into analog using the proper conversion processes and reconstruction filters should be very close to the original. If there were amplitude variations inside the audio band, then they would count as distortion, and so would prevent you from getting the excellent THD measurements that most good quality DACs can deliver. Sometimes you will see amplitude modulations at high frequencies but, as long as they occur outside the audio band, then they are... inaudible. Most DACs do produce "errors" of some sort on transient type signals (which are NOT continuous sine waves), but they tend to be minimal and not all that audibly annoying. (The variations in these "errors" account for the fact that different oversampling filters often sound subtly different, even though they may all have very low steady-state sine wave distortion figures.) However, as has been noted, there isn't a microphone, or a phono cartridge, or a speaker that is able to accurately reproduce transients anyway, so these minor errors are just one of many minor imperfections in the recording and playback chain.
    arnaud likes this.
  10. azteca x
    Keith, I'm with you. I'm a big hi-res fan and while I do think 24/96 is plenty it's up to hardware and software makers to make sure that 192 support is just fine. Though I really hope we don't get even bigger file formats. I don't take that Lavry paper as gospel at all, just offering a viewpoint. I have seen HD downloads that are just straight unfiltered DSD, presumably from SACD masters and while I haven't had any issues it is just poor practice. I'm an audio engineer myself and my preference is generally for whatever format it was recorded in to begin with even though sample rate converters are very very good these days ( src.infinitewave.ca ).
    Anyway, onward.
  11. purrin
    LMAO, I spit my morning coffee on my screen.
  12. goobicii
    20.000hz sine at 96khz,more than double the CD rate,many people say 44.1 is enough and 96khz is ultimate hi rez and than more is stupid
    only one cycle out of 5 hit the 100% volume that its supposed,the two lowest cycles are at 86% power,lets say you listen music at 100db peak volume,this means that sine would jump up and down 14db,it would only hit correct volume once in every five cycles...but wait theres more! luckily DACs dont respond instantly and dont follow the samplerate perfectly,they have settling time etc,this gives it rounded edges and actualy makes it closer to sinewave in shape becose this low samplerate draws this sine more in sawtooth shape with sharp edges resulting in distortion..... not only cant 96khz play constant sound to the point its embarassing,it cant even give shape to proper round sine,but but but muh Nyquist but Lavry said... hahaha [​IMG]
    enter the future,1536khz,perfect amplitude of EVERY single cycle within human hearing range,perfect waveform shape.Oversampling? anti aliasing filters? intersample peaks? lol
  13. Sapientiam
    What's doing the alteration? The original signal has to be band-limited before it reaches the ADC, so if you're listening with no reconstruction filter on a NOS DAC, its altered from the original.
    Incidentally you're reading too much into those Audacity waveforms - they're just points joined up with linear interpolation, whereas to get an accurate measure of amplitude you'd need to use sinc interpolation.
    arnaud likes this.
  14. KeithEmo
    I'm a little confused by these here pictures.
    If I sample a 20 Hz sine wave at a 44k sample rate, then each cycle of the wave is represented by more than 2000 sample points - which is going to deliver a pretty darned accurate "connect the dots" drawing. (That top picture looks more like a 10 kHz sine wave, sampled at 44k, then "drawn" without the proper filtering applied.)
    But what if it turned out that our digital recording did in fact reproduce a 12 kHz sine wave with all sorts of nasty jagged edges, sharp angles, and other distortions - and we didn't filter them out like we're supposed to? What would that sound like?
    Hmmmmm.... Well, since the second harmonic of 12 kHz is 24 kHz, which is inaudible to human beings, the rattiest looking 12 kHz sine wave imaginable would sound....... exactly like a perfectly clean one. (All we humans could hear would be the 12 kHz primary tone.)
    However, once we apply the proper reconstruction filter to the output, which filters out the higher frequency junk anyway, what we'll be left with is a 12 kHz sine wave....
  15. Defiant00
    As others have pointed out, what Audacity is showing you in your screenshots isn't what a DAC is going to output.
    You may find this video (with actual examples using an all-analog scope, even) informative: http://xiph.org/video/vid2.shtml
    smitty1110 likes this.
371 372 373 374 375 376 377 378 379 380
382 383 384 385 386 387 388 389 390 391

Share This Page