Home / FAQ / News
Classifieds / Events

Audio Asylum Thread Printer

Get a view of an entire thread on one page

Hi-Rez Highway

Re: SACD equilivalent sampling rates

196.2.56.5

[ Follow Ups ] [ Post Followup ] [ Display Thread ] [ Hi-Rez Highway ] [ FAQ ]

Rodney Gold

(i)

In Reply to: Re: SACD equilivalent sampling rates posted by Mars on January 20, 2001 at 22:31:14:

Yes , resolution due to noise shaping is very much increased at lower freqs , to a max of about 20 bits or so , what happens is that the noise over the whole spectrum is the same , but it is shifted to freqs or areas that are not at all critical - as you rise up the freq scale , noise increases , at about 100k , the noise is only -70 to -60db down from the signal or so. Every 6 db = 1 bit resolution , so at 100k , the SACD achieves about a 10-12 bit resolution (sort of)
Sonics will change according to what filter one uses to noise shape and what your equipment does re HF.
One must realise that we are not talking absolute values here , but incrementations and one cant really apply my prior explanation in the strictest terms.
One must also realise that DSD and PCM are not separate processes , its really just different implementations of the same type of process, neither is "better" and neither is Evil- both have compromises associated with them
Simply put , SACD is a sort of "raw" output of a 1 bit AD system that is nosise shaped and stored , conventional CDs use this Raw output but store this in another way.
SACD/DSD (whatever) is like getting the raw meat , one can do with it what one wants , mince it , fry it etc.
The other way , there are less options , the meat is already minced (bad choice of words , like the word decimation , implys some ruinous type of processing which isnt really the case) and one cant get back to the original steak to process it differently.
Thats why the Sony system is a great archival medium , it allows the user more flexibility as to what to do to the signal and assumes the least intervention possible with that signal
The "simple" analog filter Sony uses to process its data is a bit of a misnomer , the filter isnt THAT simple at all.

Rodney Gold

Follow Ups:

Re: SACD equilivalent sampling rates Mars 23:53:14 01/20/01 (3)
Thank you for your response.
I was still wondering though, whether SACD has one extra bit of resolution available at half the frequency?
You said that there are 12 or so bits of resolution at 100kHz. Does this mean that there will be 13 bits at 50kHz, 14 bits at 25kHz, 15 bits at 12kHz, 16 bits at 6kHz ect.?
If so, then this seems to be a big step down from Professional DSD which, please correct me if I am wrong, I believe to be 8 bits at 5.6kHz.
In that case perhaps a full 24 bits at all frequencies from PCM at 192kHz might sound better if the ringing of the interpolator is inaudible, and the dithering hides the errors. It this line of thinking correct?
Best regards,
Mars
- Re: SACD equilivalent sampling rates Rodney Gold 07:04:25 01/21/01 (2)
  Essentially the noise shaping curve will "set" the resolution at certian points , in that the noise will increase , as noise increases one loses resolution in terms of the softest vs loudest sound that can be heard.
  So the answer to your question re res drops no's is not certian unless one has access to the curves , which I dont have: If the SCD-1 will allow a 100khz signal to be passed , as it does , then the noise there cannot be that excessive.
  SACD will attain max resolution to well over 20khz and it will reduce according to the curve , your speakers wont do 100khz and neither will you hear it apart from there being little or no info on cd up to there as well as the fact that few music instruments wll reach anywhere near there , the output at 100khz is so low it is sort of academic.
  There are studies which seem to prove that even tho we dont hear after 20khz , the signal still influences us.
  As to your 24/192 Q
  Errors at 24/192 are actually negligable compared to 16/44 , of the order of 256x less!!!!
  One doesnt need to apply dither if one is not truncating , and at 24/192 there is no truncation , oversampling and the fact that the anti aliasing filter is at 90+khz , means that the filters employed in 24/192 vs 16/44 are FAR more benign and gentle.
  All in all , on both sides , the AD and the DA process , 24/192 is very gently acted on and will preserve the signal most likely the same as a DSD type process will.
  No one has compared the same music/recording/mix via both processes , so one cant say whether the "pcm" method or the DSD method is better. Time will tell whichh remains dominant
  
  Rodney Gold
  - Re: SACD equilivalent sampling rates mattpeterson 22:07:34 01/21/01 (0)
    Rodney Gold wrote:
    All in all , on both sides , the AD and the DA process , 24/192 is very gently acted on and will preserve the signal most likely the same as a DSD type process will.
    No one has compared the same music/recording/mix via both processes , so one cant say whether the "pcm" method or the DSD method is better. Time will tell whichh remains dominant
    ------
    This is true. Until the same performance with the same mic feeds is coded to both DSD and 192kHz simultaneously, and this is offered to end consumers in both formats, there will be no way for the average audiophile to compare the two.
    But, DVD-Audio does not have room for more than two channels at 192kHz. SACD can support up to six 64fs DSD channels. This apparent deficiency with DVD-Audio will probably be offset in the mind of the public by longer playing times available with MLP, but only SACD (in current configurations) is capable of full high resolution on all surround channels.
  - Re: SACD equilivalent sampling rates Mars 10:19:49 01/21/01 (0)
    When trying to understand the SACD noise shaping filter, I find it more intuitive to imagine what the output voltage will do for each new clock cycle. Perhaps this is too reductionist. I shouldn't have used the word "simple" as I really don't understand the details. Can you explain why it is said to "shift the noise?" This is not intuitive enough for me.
    Also, as for PCM 24/192, will the anti-aliasing filter of 96kHz add undesirable phase distortion presumably not present in SACD?
    Thanks in advance,
    Mars
Re: SACD equilivalent sampling rates Jazz Inmate 23:06:03 01/20/01 (0)
> > One must also realise that DSD and PCM are not separate processes , its really just different implementations of the same type of process, neither is "better" and neither is Evil- both have compromises associated with them < <
I realize you are speaking theoretically here, but in implementation it seems to some of us that PCM has a real-world sonic signature associated with it that ranges from the subtle/insidious to the overtly harsh. . .while this may not be evil, it is certainly not desireable and as you say, DSD is a more flexible medium. even from a theoretical standpoint, is DSD not "better" by this virtue alone? and if so, why continue developing PCM at all?

This post is made possible by the generous support of people like you and our sponsors:

Joseph Audio

You can not post to an archived thread.