[riverlethe]

This is an earlier study showing no discernible differences between DVD-A (PCM) and SACD (DSD).

http://issuu.com/manger-msw/docs/aes_paper_6086/1

 

[miceblue]

That's between DSD and down-sampled PCM though. It's not exactly the a fair comparison to say higher sampling rates provide no audible benefits. XD

The article I linked to talks about a comparison between SACD and DVD-a vs those same sources down sampled through an analog-digital-analog loop. The 554 trials they performed demonstrate that it's a perfectly fair comparison.

DSD is 1-bit, some megahertz sampling. That's not even close to 24-bit, 192 kHz sampling rates.

DSD vs PCM is a completely different argument from PCM vs PCM or even DXD.

 

[ralphp@optonline]

DSD is 1-bit, some megahertz sampling. That's not even close to 24-bit, 192 kHz sampling rates.

DSD vs PCM is a completely different argument from PCM vs PCM or even DXD.

Despite the big DSD push from several high end audio manufacturers and the high end audio press, DSD is NEVER going to catch on because of the limitations imposed by DSD when it comes to editing and production. In order to edit a DSD recording one must convert the DSD files to PCM, edit and then convert the PCM files back to DSD. So other than a pure, unedited DSD recording ALL DSD recordings are, in essence, PCM.
 
So comparing DSD to PCM is totally fair.

 

[riverlethe]

DSD is 1-bit, some megahertz sampling. That's not even close to 24-bit, 192 kHz sampling rates.

DSD vs PCM is a completely different argument from PCM vs PCM or even DXD.

The differences are moot if they're inaudible.

 

[Don Hills]

The differences are moot if they're inaudible.

Finally some sanity.

 

[Brooko]

As far analog signal has infinite possibilities
You can never code it effective with any amount of buts

 
 

It doesnt matter what dynamic range are you coding
More bits mean fine step ups and downs
And thats on xyz

I know.what do u mewn with that
And that happens very often
You are right in most way
But......16 bit steps arent fine enough for some worst case scenarios
Its not accident they choose 16 bit
It is good in most situationts snd for nearly everybody

16 bit has worse steps as i said
It means you need lil bit more capacity infouence to clean the stepy wave to wavy signal
More bit means less step level so less infouence on compensating
Its not some audio tech but just digital basics

 
Actually you couldn't be more wrong.  I'm guessing you also didn't follow that first set of links I left for you either 

 
If you don't follow any of the other links - just please follow this one - it is a video but worth it.  It explains directly why what you posted is a common fallacy - http://xiph.org/video/vid2.shtml
 
And here's some commentary on Neil Young's Pono which has direct commentary on HD audio (what we can and can't hear).  Again - you may find it enlightening ......
http://news.cnet.com/8301-1023_3-57620489-93/sound-bite-despite-ponos-promise-experts-pan-hd-audio/

 

[Digitalchkn]

 

  Actually you couldn't be more wrong.

 
 
 
 

   
 
 
Actually you couldn't be more wrong.  I'm guessing you also didn't follow that first set of links I left for you either 

 
If you don't follow any of the other links - just please follow this one - it is a video but worth it.  It explains directly why what you posted is a common fallacy - http://xiph.org/video/vid2.shtml
 

I encourage folks arguing either side of the "analog vs digital" to refer to chapters 2 and 3  of Digital Audio Signal Processing by Udo Zolzer  for an excellent description of the sampling, reconstruction, quantization, and noise shaping theory, before accepting watered-down versions suggested on various youtube videos as the gospel. As usual, there is more than meets the eye.

 

[Brooko]

I encourage folks arguing either side of the "analog vs digital" to refer to chapters 2 and 3  of Digital Audio Signal Processing by Udo Zolzer  for an excellent description of the sampling, reconstruction, quantization, and noise shaping theory, before accepting watered-down versions suggested on various youtube videos as the gospel. As usual, there is more than meets the eye.

Funnily enough I downloaded and started reading the book. Basically got lost within the first few pages. You realise it's basically an advanced text book right? Not exactly user friendly to us hobbyists. Personally I'll stick to Montgomery's watered down version. At least I could understand that one. Thanks for the reference though.

 

[Digitalchkn]

Funnily enough I downloaded and started reading the book. Basically got lost within the first few pages. You realise it's basically an advanced text book right? Not exactly user friendly to us hobbyists. Personally I'll stick to Montgomery's watered down version. At least I could understand that one. Thanks for the reference though.

Well, I never said it is light reading

 
If you do go through it, one day, you'll find lots of fascinating things as a result. Some examples of this are:
- Your ability to get a *perfect* copy of analog signal is to have a theoretically ideal reconstruction filter .... that is basically impossible to create in a real "PCM" DAC.  Fortunately, these days there are tricks to overcome this, but unfortunately it sometimes makes the direct A/B comparisons kinda to hard to make (particularly when using different dacs).
- SNR is actually not a single number (e.g. 96dB for 16 bits) but highly depends on the "randomness" of the signal you are trying to sample
- We can apply a mathematical "trick", referred to as Dithering, that takes advantage of limitations of our hearing to make it seem like we are using more bits. The effectiveness of this depends on the type of dithering that is being applied and, again, the input signal itself.
- Analog tape noise (for instance) is not same *type* of noise as sampling noise, making things a bit more difficult to make apples-to-apples comparisons. In fact, digital noise changes depending on the input signal.
 
That's why the watered-down version should be treated with care!

 

[Brooko]

Again though - in the "watered down version" and in his blog, Monty does list exactly those sort of things that we need to take care with (ie use of dither, background noise, theoretical noise floor etc, etc).  But what I like about his video and blog is that they are easier for a 'layman' (translate to idiot

) like me to understand + he also applies it to what is real-world (ie audible).
 
I think that Monty and (the little I understood of) Zolzer were in agreement though - it is possible to digitise a wave form so that any difference is inaudible.  That's where the rubber really hits the road (so to speak).
 
Curious DC - what field are you in?  You seem very knowledgeable on the topic.  I'm enjoying your insights.

 

[Digitalchkn]

  Again though - in the "watered down version" and in his blog, Monty does list exactly those sort of things that we need to take care with (ie use of dither, background noise, theoretical noise floor etc, etc).  But what I like about his video and blog is that they are easier for a 'layman' (translate to idiot

) like me to understand + he also applies it to what is real-world (ie audible).
 
I think that Monty and (the little I understood of) Zolzer were in agreement though - it is possible to digitise a wave form so that any difference is inaudible.  That's where the rubber really hits the road (so to speak).
 
Curious DC - what field are you in?  You seem very knowledgeable on the topic.  I'm enjoying your insights.
 

 
I am neither in the music business nor in hi-fi equipment.  I am an electrical engineer in a field completely unrelated to audio. So this is purely self-interest. Been interested in sound recording, reproduction since I was a teenager.  Amateur guitarist (for fun). Occasionally make recordings in spare time in a home environment. Hence, the rate conversion software.
 
Don't feel like you are left out. I don't think it's light reading for anyone, really.  It does help to have an electronics engineering degree with some signal processing focus to help dive deeper into nuts and bolts of it  (I don't use it much of it these days so am a bit rusty on the math).

 

[miceblue]

I encourage folks arguing either side of the "analog vs digital" to refer to chapters 2 and 3  of Digital Audio Signal Processing by Udo Zolzer  for an excellent description of the sampling, reconstruction, quantization, and noise shaping theory, before accepting watered-down versions suggested on various youtube videos as the gospel. As usual, there is more than meets the eye.

Interesting. I'll have to read through this now that I'm on spring break.

I'm studying bioengineering, though far away from the tech stuff. I'm more into the biological side of it with tissue engineering and whatnot. I did take a few courses regarding signal processing though since 1) this hobby, 2) it might be useful for understanding cochlear implants (though again, I would be more interested in a tissue engineering approach to restore hearing).

 

[stv014]

Funnily enough I downloaded and started reading the book. Basically got lost within the first few pages.

 
Which is probably why it was posted. It is a common tactic by audiophiles to try to confuse people with walls of information, and use the "audio is infinitely complex and you do not understand it" argument to justify believing in whatever they want.

 

[stv014]

Originally Posted by Digitalchkn /img/forum/go_quote.gif
 
In fact, digital noise changes depending on the input signal.

 
Well, with dithering (the simple +/-1 LSB TPDF type), I can subtract the quantized signal from its original version, and the difference (the quantization error) sounds and looks (in a spectrum analyzer) like white noise at a constant RMS level no matter what the input signal is. This is easy to verify in practice.

 

[stv014]

Originally Posted by Digitalchkn /img/forum/go_quote.gif
 
- We can apply a mathematical "trick", referred to as Dithering, that takes advantage of limitations of our hearing to make it seem like we are using more bits. The effectiveness of this depends on the type of dithering that is being applied and, again, the input signal itself.

 
It is actually noise shaping that takes advantage of the limitations of hearing (notably the hearing threshold increasing significantly towards 20 kHz) to improve the perceived dynamic range of quantized audio. Simple dithering sounds like white noise, or just slightly colored to reduce the A-weighted noise level by 1-2 dB. Also, for correctly implemented dithering, the input signal should not matter, other than the theoretical possibility of it correlating with the PRNG used for dithering, which in practice should be negligible. Of course, a louder input will perceptually mask the noise more, but that is the same for analog noise as well.