avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#41 [url]

Dec 13 16 11:53 AM

morespaceecho wrote:
zmix wrote:
Dithering 24 bit files or exporting 32 bit floating point files is an "it couldn't hurt"  practice firmly founded in superstition, not science.

 

i'm sorry for making you sad, but i can't help it:

we have a track with a bunch of plugins on it. or just one, whatever. a 32fp render of this track will be exactly the same as the source. this is easily verified with a null test. a 24 bit render will be almost exactly the same, except there'll be either truncation distortion or dither noise added. also easily verified with a null test.

how is this anything other than straight up rudimentary bulletproof science? there's nothing superstitious about it. saying "it's happening at a level that's far too low to hear" is fine and i agree with you, but that's like telling a scientist "oh please, if you need a microscope to see it, it can't possibly be important."
 
First of all, let's be perfectly clear about what you're claiming.

You state that a 32 bit float file represents the source better than a 24 bit file. 

It also appears that you've modified the source with DSP operations and you've stated previously that you believe these operations expand the word width to 32 bits.

Are you comparing the  24 bit source to the DSP modified file in this case?

If so it appears that you're also asserting that the truncation to 24 bits has compromised  the data integrity of the now 32 bit file.

Do I understand you correctly?

Now lets add some missing details:

A 32 bit floating point file contains 24 bits of audio data and an 8 bit multiplier, which allows the level to be scaled.  If the 24 bit audio data is extracted from the file, it is identical to a normalized 24 bit file.

When a 32 bit float file is reproduced it is only ever a scaled 24 bit file.

Dither and truncation should only happen when the output file is being generated. 

For these reasons, comparing the dithered 24 bit file to an internal DAW file format is irrelevant.

There is no output mechanism on earth that can reproduce the quantization error or noise of a truncated 24 bit signal.

And lastly, as demonstrated in the posted video, and I'm talking about the part where he's normalizing the gain for the truncation and performing a null test using the plugin he created, the self noise of a file is so far above the quantization error that the error becomes immeasurable.


It's important to remember that each additional bit only adds another least significant bit to the word width, and has no affect on the upper bits.


--


 



Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#43 [url]

Dec 13 16 1:12 PM

morespaceecho wrote:
i'm happy to argue this for another 10 pages, but i dunno that anyone feels like reading it, so how about we just agree to disagree, as gentlemen do.

It's not a difference of opinion...

Quote    Reply   
avatar

maarvold

Aqua Marine

Posts: 3,205 Member Since:23/01/2011

#45 [url]

Dec 13 16 1:35 PM

zmix wrote:

maarvold wrote:
gtoledo3 wrote:Why not have a two car garage for your one car? Why not a four car garage for that matter...

 Even though we have only 2 cars, a 4 car garage would give me space for a workbench and I could actually take my bike from my back patio and store it in the garage... it would also give my wife room for storing items bought in bulk from Costco (something she's been wanting).  But I digress...

None of which affects the capacity of the car or it's ability to transport things or people.

Dithering 24 bit files or exporting 32 bit floating point files is an "it couldn't hurt"  practice firmly founded in superstition, not science.

...

 
Hey Chuck--believe it or not, my comment was ONLY about what I might do with a 4 car garage--it probably belonged in the 'staying out of the doghouse' thread.  Sorry for the hijack.  And, for the record, a 4 car garage would: A. never fit on my property, B. be totally out of balance with the scale of my house.  And now I am really digressing...  P.S., These comments are also ONLY about my garage.  

Quote    Reply   
avatar

morespaceecho

Platinum Blonde

Posts: 2,323 Member Since:29/01/2011

#46 [url]

Dec 13 16 1:35 PM

right, sorry. 

original source is a 24 bit file. if i do nothing at all to it and render a 24 bit copy, it will indeed be 100% exactly the same.

if i process that original source in any way, the word length expands (easily confirmed by looking at a bit scope). if i then render that process to a new file, a 32fp file will be exactly the same, a 24 bit file will either have dither noise or truncation distortion added. yes it is at a level only hearable by specially-trained elite bats, but it is there. 

anyone can test this for themselves in less time than it takes to reply telling me why i'm wrong. 

www.oldcolonymastering.com

morespaceecho.bandcamp.com

Quote    Reply   
avatar

soapfoot

Ruby Baby

Posts: 7,512 Member Since:04/02/2011

#47 [url]

Dec 13 16 2:08 PM

zmix wrote:

soapfoot wrote:...and I certainly have my own superstitions when it comes to audio (and other things), so who am I to judge?

Online discussions about digital audio make me so sad....

We're living in an era of anti intellectualism, social media has given everybody the same platform, which they can use  to originate or regurgitate information.

It's a gish gallop of solopsism and emotionally charged rhetoric, exactly paralleled by the ubiquity of music making tools and the resultant conformity.

People who should know better shamelessly peddle half truths because it serves their business interests.

Observations and practices based on superstition,  gut feelings,  or anecdotal evidence are not facts, and one should pay close attention when the discussions about things they clearly don't understand make them feel upset,  because therein lies madness.


It becomes tricky when the "artist" and "engineer" roles are occupied by the same person, as they so often are now. Solipsism on the part of an artist isn't always a bad thing. On the part of an engineer, it's not so good. Same with superstition.

I understand and relate to the angst (I have my pet peeves, too... like when a guitarist uses "gain" as a synonym for "distortion"). 

But really, whatever makes someone feel secure in their creative space (therefore allowing them to do their best work) and isn't hurting anything is OK by me (even if it's absurd objectively). 

 

brad allen williams

Last Edited By: soapfoot Dec 13 16 2:12 PM. Edited 1 time.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#48 [url]

Dec 13 16 2:55 PM

tom eaton wrote:
Chuck, he's saying that a 32 bit fp capture of a process maintains the result of the process more completely than a 24 bit capture. He should not have used the word "source" as you read that to mean the original unprocessed "source file."
The claim is speculation.

The audio file within the 32 bit file is still only 24bits wide.

It doesn't matter how many cars your garage can fit, when you leave the garage, you're still only driving one car.


REMEMBER:  This discussion is about dither. Dither only applies to final analog output files.  All DAW environments use 32 bit or higher processing, and comparing a dithered 24 bit output file to an internal DAW file format is irrelevant.




Quote    Reply   
avatar

dcollins

Platinum Blonde

Posts: 2,373 Member Since:27/01/2011

#50 [url]

Dec 13 16 3:58 PM

minister wrote:
zmix wrote:

soapfoot wrote:...and I certainly have my own superstitions when it comes to audio (and other things), so who am I to judge?

Online discussions about digital audio make me so sad....

We're living in an era of anti intellectualism, social media has given everybody the same platform, which they can use  to originate or regurgitate information.

It's a gish gallop of solopsism and emotionally charged rhetoric, exactly paralleled by the ubiquity of music making tools and the resultant conformity.

People who should know better shamelessly peddle half truths because it serves their business interests.

Observations and practices based on superstition,  gut feelings,  or anecdotal evidence are not facts, and one should pay close attention when the discussions about things they clearly don't understand make them feel upset,  because therein lies madness.

Then agin, there are those who learn from those and these discussions who mightn't not otherwise...................................................
 
For instance, you can’t really d*ther floating point.  Maybe you can get a result for some words that is slightly better, maybe slightly worse, but it’s nothing like the perfect effect you get from doing it to fixed-point.

Quote    Reply   
avatar

gtoledo3

Aqua Marine

Posts: 4,237 Member Since:23/10/2013

#51 [url]

Dec 13 16 4:18 PM

What happens in PT, Logic, or any mix engine that calculates at something like 56 or 64 floating point for fader adjustments, but outputs a 32 bit float buffer from plugin to plugin? Wouldn't TPDF be applied at some point internally after the higher bit depth processing?

Last Edited By: gtoledo3 Dec 13 16 4:21 PM. Edited 2 times.

Quote    Reply   
avatar

tom eaton

Platinum Blonde

Posts: 1,335 Member Since:25/01/2011

#52 [url]

Dec 13 16 9:33 PM

 

Dither only applies to final analog output files

 

I like this bit from Paul Frindle:

"any undithered digital representation of an audio signal is effectively illegal."

That's from the Oxford Limiter section on dither, which is particularly good at laying out the principles at play.

And it really has nothing to do with analog at all.

Last Edited By: tom eaton Dec 13 16 9:36 PM. Edited 1 time.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#53 [url]

Dec 13 16 10:21 PM

tom eaton wrote:
 

Dither only applies to final analog output files

 

I like this bit from Paul Frindle:

"any undithered digital representation of an audio signal is effectively illegal."

That's from the Oxford Limiter section on dither, which is particularly good at laying out the principles at play.

And it really has nothing to do with analog at all.
Tom,

I don't think you've understood the examples in the Sony Oxford Limiter manual.  

They are AP plots showing the result of truncation to 16 bits with and without dither.

Here is the full text of the very small part you quoted:

The graph below illustrates this fact by showing a 1KHz signal at —120 dBFS passing through a dithered 16 bit system. This corresponds to a signal 24 dB below the level of the least significant bit; the effective channel SNR (Signal to Noise Ratio) is added in blue for illustration purposes.

This illustrates that dither turns a quantised numerical signal conduit into the equivalent of a naturally continuous (un-quantised) system, which exhibits a finite signal to noise ratio with no practical limit to harmonic signal resolution. In other words the inescapable presence of quantisation in numerical systems does not forcibly lead to ‘discontinuity’ or ‘resolution loss’ in the signal. The misunderstanding of this fact underpins many of the most damaging misconceptions surrounding digital audio systems. It can also be deduced from the above graphs that any undithered digital representation of an audio signal is effectively illegal.


Also, from the same manual, you might find this instructive:

9.4 Additional Information

Generally speaking, a signal must be dithered if it is to be output via any form of reduced bit-depth signal (for example, you are listening to it via a DAC) or to any limited resolution media. The only time you can avoid dithering your material is if you know for certain that dithering will be applied later (for example your material will be sent to a mastering studio).

In fact, one might accurately infer from this that a 32 bit float file might be "legal" for use as an interim file type,  but only as such, since it's not a final output format, which obviates the claim of 32 bit float being  more true to the source than a 24 bit file...



Last Edited By: zmix Dec 13 16 10:26 PM. Edited 1 time.

Quote    Reply   
avatar

tom eaton

Platinum Blonde

Posts: 1,335 Member Since:25/01/2011

#54 [url]

Dec 13 16 10:38 PM

Chuck, I understand the issue quite well, thanks! You posted that dither was for analog files. I thought I'd add some facts. Paul seemed a pretty good source and I'm glad you agree with him. You'll see that nothing I stated is contrary to his thinking in any way.

Not all of the people on the internet need to be spoken down to, or presumed to be ill informed.

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#57 [url]

Dec 14 16 9:03 AM

tom eaton wrote:
Chuck, I understand the issue quite well, thanks! You posted that dither was for analog files. I thought I'd add some facts. Paul seemed a pretty good source and I'm glad you agree with him. You'll see that nothing I stated is contrary to his thinking in any way.

Not all of the people on the internet need to be spoken down to, or presumed to be ill informed.

It's not my intention to do either, Tom, but when you make statements like "It has nothing to do with analog at all"  I  start to think you are misunderstanding the source you're citing, or partially quoting the source to make a point not in evidence in the full text.


Quote    Reply   
avatar

jerry tubb

Gold Finger

Posts: 659 Member Since:06/02/2011

#58 [url]

Dec 14 16 9:25 AM

Ha! The most excitement this forum has seen in months!
It's all in a Dither about Dither.
It's just Dither and bit depth reduction, not life vs death.
I think we all studied this stuff years ago, got a basic understanding of it, then use it while we're working, often automagically.
And we try not to overdo the Dither, e.g. Multiple layers of noise shaped Dither on 16-bit files etc...
Fair enough?
Best regards, JT

Quote    Reply   
avatar

zmix

Aqua Marine

Posts: 4,285 Member Since:20/01/2011

#59 [url]

Dec 14 16 9:31 AM

And now, for a complete change of pace, I'll disclose my own personal OPINION about d*ther:

If people want to dither 24 bit files, have at it.  If people want to send me 16 bit undithered files for mixing, that's fine too....

However, I do want to state my OPINION about "noise shaped dither".

 Listening to audio processed with Noise Shaped Dither gives me a headache.

Noise shaped dither was an answer to a question nobody was asking, except the bit counters in the labs... (reminder, OPINION..!!!!)

I have always been able to pick it out..  I suspect I'm not alone in this opinion.  Several artists and producers over the years have lamented to me that "The CD sounds so much worse that what we heard in mastering"...    Back in the 1890s, in the early days of noise shaped dither, I was asked by a producer to listen to the ref CD and decide if we should remix or at least remaster the record.  The symptom was that all the tambos, cymbals, fricatives and siblants  sounded horrible.. very "digital"...  We compared the DAT mixes (1890s, remember?) they sounded fine.  I had not attended the original mastering session, so we decided that I should talk to the mastering engineer.. immediately upon arrival I spied an Apogee "UV22" converter.     "UV22" was a form of noise shaped dither that concentrated the energy of the d*ther at nyquist.    I asked the mastering engineer if he had used UV22 on the master and he smiled and said "Yep, it's the best..!"  I suggested that I'd like to hear the same treatment without the UV22, as it was *new* and *the best* he had that familiar dogmatic reluctance to accommodate that so often plagues the superstitious, but eventually agreed to "humor" me...   When we ran the master again without UV22... all of the issues we heard were gone..  I remember pointing out  that the fidelity was so much improved that they might sell more units if they wired the UV22 switch backwards...



 


Quote    Reply   
Add Reply

Quick Reply

bbcode help