The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Good dither practices, what are yours? Dynamics Plugins
Old 10th January 2017
  #61
Lives for gear
 
toledo3's Avatar
 

Quote:
Originally Posted by bcwiz View Post
I've read a lot of discussions on dither and I'm still confused. I'm running Pro Tools native 24/48k (so floating point internal). I typically print stems in the same session, drums, bass, strings, etc. I take from these discussions that I should have a dither plug on the master before each of those print tracks and on the master that's feeding the DAC going to the monitors. Is that correct? Some say yes some say no. Which is it? The only dither plugs that come with Pro Tools have settings for 8,16 and 20 bit, no 24 bit. Why no 24 bit?
If this is a necessary part of using a floating point DAW then why the hell isn't the dither built into the damn thing in the first place, or is it?
I've only dithered when going to 16 bit.
Because you don't need to out think what the DAW mixer does behind the scenes to ensure your 24 bit dynamic range. That is why there is no 24 bit dither.

There is absolutely NO expectation that when working in a 24 bit session, and the DAW internally steps up to say, 32 bit, that it will then do the wrong thing and truncate it on the way out of the DSP chain, and that you need to dither it.

As a programmer, I find it really weird when I see companies get voodoo about this...
Old 10th January 2017
  #62
Motown legend
 
Bob Olhsson's Avatar
 

There are plug-ins that tell you the bit depth. Samplitude defaults to properly dithering and has "smart" dither. This is a whole bunch of why it "sounds better." Logic offers an option to dither 24 bit mixes (there is no floating point option) but every choice is some form of noise shaping, a left over from twenty years ago.

We had to fight with Avid to get them to include a dithered mixer plug-in for the TDM system which truncated every single feed into a plug-in. The included plug-ins in Pro Tools are legacy from that system that dithered everything to 24 bits if you knew to substitute the dithered mixer in the folder. That's important for backward compatibility but there really ought to simply be a built in TPDF choice.
Old 10th January 2017
  #63
Lives for gear
 

Interesting stuff!
Old 10th January 2017
  #64
Gear Guru
 
Karloff70's Avatar
 

Interesting indeed. I had a wee hunch dither being dealt with properly would be a DAW dividing sort of issue that might have something to do with people hearing them sound different.......

Thank you for frequenting this forum, Mr Olhsson!

*sitting back and taking in the flow of goodness*
Old 11th January 2017
  #65
Lives for gear
 
Brian Campbell's Avatar
 

Quote:
Originally Posted by toledo3 View Post
Because you don't need to out think what the DAW mixer does behind the scenes to ensure your 24 bit dynamic range. That is why there is no 24 bit dither.

There is absolutely NO expectation that when working in a 24 bit session, and the DAW internally steps up to say, 32 bit, that it will then do the wrong thing and truncate it on the way out of the DSP chain, and that you need to dither it.

As a programmer, I find it really weird when I see companies get voodoo about this...
So you're saying there's no need to dither when recording stems to tracks in the same session when running 24 bit in a native rig.
Also interesting to note that according to RME the Totalmix FX dsp mixer is not dithered.
Old 11th January 2017
  #66
Lives for gear
 
Fidelis's Avatar
 

Hi Bob, I use to use the dithered mixer on my TDM system. But I got the impression that HDX and HDN have the dithered mixer built in, am I wrong?

So, for mixing on a console this days, do you think every output have to be dithered to 24bit? I keep my sessions on HDX at 24bit, not 32. Even this way, if I process any track with a plugin or gain chance, does it goes to 32bit so I would have to dither it before sending out to the console?

Thanks for all your input.

Quote:
Originally Posted by Bob Olhsson View Post

We had to fight with Avid to get them to include a dithered mixer plug-in for the TDM system which truncated every single feed into a plug-in. The included plug-ins in Pro Tools are legacy from that system that dithered everything to 24 bits if you knew to substitute the dithered mixer in the folder. That's important for backward compatibility but there really ought to simply be a built in TPDF choice.
Old 11th January 2017
  #67
Motown legend
 
Bob Olhsson's Avatar
 

I honestly believe lack of dither and poorly designed analog stages in converters is most of why people don't like the 'sound" of digital recordings and think analog is magical.
Old 12th January 2017
  #68
Lives for gear
 

craigb, I'm a bit confused on your monitoring settings. RME's software mixer should not produce truncate error noise unless you have it change gain or mix signals so 24 bit into the RME should be the same 24 bit out of the RME going to whatever DAC set to 24 bit (further link shows bit-accuracy at unity gain, i. e. it passes 24-Bit CRC tests). Even with gain change / mixing in the RME mixer it's questionable you are going to get the amount of error noise you do in the DAW app itself or that it would be worth applying a dither plug to it's mixer's ouputs. Link: https://www.rme-audio.de/en/support/...x_hardware.php


There is also the debate on WHEN or IF to dither if sending off for mastering as the effects of cummulative dither can be a very bad thing too. Links:

Dither - Digital Domain: CD Mastering | Mastered for iTunes | Audio Mastering | Blu-Ray Mastering
Dithering Explained: What it is, When to Use It, and Why it’s Important

My preference is to use dithering only at the last stage of the DAW if converting down to a 16 bit output. If further processing by others / mastering, then leave DAW ouput at 24 bit fixed and skip dithering. If you want to monitor in 16 bits with dither for checking there are several ways you can configure for a quick change or simply print two separate stereo files (one undithered 24 bit and one dithered 16 bit). Printed 24 bit stems will be undithered as they will be further processed if needed by myself or others.

On the whole debate of a project setting of 24 bit fixed or 32 bit float . . . the DAW is going to do it's internal thing at whatever and the gain safety net of a project at 32 bit float is not worth the disk drive performance penalty to me. We've had a few threads on that subject here too LOL.

I guess the biggest debate this thread opens up is whether cummulative 24 bit dither is more desirable than cummulative 24 bit truncate error noise? Probably either is not going to be heard but for Classical recordings. Maybe others will chime in on this issue.
Old 12th January 2017
  #69
Lives for gear
 
JulenJVM's Avatar
Quote:
Originally Posted by Bassmankr View Post
I guess the biggest debate this thread opens up is whether cummulative 24 bit dither is more desirable than cummulative 24 bit truncate error noise? Probably either is not going to be heard but for Classical recordings. Maybe others will chime in on this issue.
I think the issue with truncation distortion is not that you might be able to hear it, but is rather about what it takes away from the mix, the finer detail that disappears and makes things sound flat, thin...and digital.

About cumulative dither, I'm personally fine with a 24 bit dither pass plus a 16 bit pass at the mastering stage. To my ears, this sounds better than not applying the 24 bit dither on the mix print, which is pretty low in volume by the way, and not perceptible to my ears even after the 16 bit pass.
I can't comment on dithering stems and its cumulative effects, but you have specialized dithers for the task, that do the dither job in a very stealthy way.
Old 12th January 2017
  #70
Motown legend
 
Bob Olhsson's Avatar
 

Truncation distortion being correlated to the audio accumulates 6 dB. a pass while random dither only accumulates 3dB. The idea of dither accumulation being a problem is nonsense when the only other choice is even more audible distortion.

Most of the information published about dither accumulation and noise shaping was written twenty years ago based on people's experiences with that era's converters. With modern converters, I've found the theoretical ideal TPDF or hpTPDF tends to sound the best. Because most people today listen to audio through digital volume controls and lossy encoding, noise shaping is not a good idea because there is almost always that additional processing.
Old 12th January 2017
  #71
Motown legend
 
Bob Olhsson's Avatar
 

As far as I know ASIO, along with the input to most converters and AES/EBU/SPDIF connections all truncate to 24 bits. This is why you need to dither the monitor feed of many DAWs. It's interesting that the original digital telephone system from the 1950s was properly dithered!
Old 12th January 2017
  #72
Lives for gear
 

So Bob, you are basically recomending using a separate (mono if not output in stereo pairs) 24 bit TPDF or hpTPDF dither plugin as last in chain for ALL hardware inserts or in the case of a hybrid mixing setup to an analog desk, a separate (mono if not output in stereo pairs) 24 bit TPDF or hpTPDF dither plugin for EACH DAC converter channel used (in my case 72 channels through three RME 9652 cards - time to make a template LOL) and for ITB mixing then just a final 2 buss 24 bit TPDF or hpTPDF dither plugin IF (take a breath) the DAW app itself does not do this smart stealth dither automatically. So your opinion is that the above mentioned dither type is the lesser of two evils regardless of number of output channels and that cummulative effects regardless of channel counts or further work on the file are today a non-issue with that 24 bit dither? If that is the case then a listing of each DAW app's functions before output to the audio interface would be the thing to know (do I need to add a plug or does the DAW app itself do that already).

Next - while I see 24 bit dither would be best to apply via the audio interface (given your ASIO info) I'm sure there is a latency / performance tradeoff with applying it there. It's already a performance latency vs. quality issue when the converter maker is picking the profile. Working in realtime is a Bitch (with small buffers at least).
Old 12th January 2017
  #73
Motown legend
 
Bob Olhsson's Avatar
 

You can only choose between dither and distortion. End of story!
Old 12th January 2017
  #74
Gear Guru
 
Karloff70's Avatar
 

Quote:
Originally Posted by Bob Olhsson View Post
You can only choose between dither and distortion. End of story!
So it's 3dB of dither noise or 6dB of truncating distortion being the choice per pass? Seems an easy choice.

You speak of ASIO, do you know if Apple's Core audio has the same issues and needs dither on the monitoring?
Old 12th January 2017
  #75
Lives for gear
 

Quote:
Originally Posted by JulenJVM
I think the issue with truncation distortion is not that you might be able to hear it, but is rather about what it takes away from the mix, the finer detail that disappears and makes things sound flat, thin...and digital.
Quote:
Originally Posted by craigb
As Julen said, it's not so much that the truncation distortion itself is audible, but that it's interaction with the actual audio is, subtle though it may be. As others have noted, Bob Katz among others recommends 24bit dithering any time you go out to hardware. I'm not doing anything in Totalmix, it's all at unity.
If a process causes distortion and you hear a difference before and after this process.. and nothing else have been changed, then it is distortion you hear.

It does not make sense to say that the distortion is inaudible, but the effect of the distortion on the audio is audible. You have one signal only. If this signal is changed by a nonlinear process enough to make an audible difference, you hear the distortion of the signal.
Old 12th January 2017
  #76
Motown legend
 
Bob Olhsson's Avatar
 

I have no idea because I dumped Apple 15 years ago.
Old 12th January 2017
  #77
Motown legend
 
Bob Olhsson's Avatar
 

The distortion may not be audible after one generation of processing but it speeds up the point where the audio becomes crunchy. You must always look at this stuff as part of a system.
Old 12th January 2017
  #78
Lives for gear
 

Quote:
Originally Posted by craigb View Post
Hi Bassmankr,
As Julen said, it's not so much that the truncation distortion itself is audible, but that it's interaction with the actual audio is, subtle though it may be. As others have noted, Bob Katz among others recommends 24bit dithering any time you go out to hardware. I'm not doing anything in Totalmix, it's all at unity.

MC of RME support says:
"Use only 24 bits. If you send 32 the driver will strip off the lower 8 automatically."
https://www.forum.rme-audio.de/viewtopic.php?id=10617

There's the truncation happening right at the ASIO driver input of Totalmix. The Reaper monitor bus is full 64bit float, confirmed with Bitter. ASIO can handle 32 bit, but RME does not, and just chops it to 24 bit.
You are mixing floating point and fixed point together. What Matthias is saying is if you send 32 bit fixed point that is what RME will do. Floating point is not directly translatable to fixed point. There is no "chopping" (truncation) of floating point -- it first gets translated to a fixed point approximation at the bit depth required.
Old 12th January 2017
  #79
Gear Addict
 
Blaine Misner's Avatar
 

Quote:
Originally Posted by JayTee4303 View Post
U87ai>AvalonM5>Input 1 on Motu HD-192. Straight into DAW, AND...

Back out Output 3 of HD-192> UA1176>LA2a>Input 5 HD-192

Straight into DAW from Input 5, AND...

Back out HD-192>Artist's headphone mixer.

Prefer the option of recording both straight, and slightly compressed audio, each take. Could do it all OTB, on the way in, but my time machine is down for repairs.

Since no dither is possible in the HD-192 DSP software, can I assume the move is to control 1176 input with the Avalon M5?

Or am I missing something, and using gain or attenuation in the HD-192 DSP software, between the M5 output and 1176 input won't contribute truncation distortion?

If not why not? (Part of a larger set of questions, where one answer applies to all.)
Dude. just mult the signal after the avalon. you're making a needless round trip through the convertor.

Last edited by Blaine Misner; 12th January 2017 at 10:48 PM.. Reason: typo
Old 13th January 2017
  #80
Lives for gear
 

Quote:
Originally Posted by craigb View Post
I understand the difference, my point is that the 64bit float internal path of Reaper is getting converted to 24bit fixed somewhere. Is it possible to convert a true 64 float down to 24 fixed without loss?
No. You can never convert fp to fixed without loss.

I also know of no floating point DAC in use. There are 32 bit fixed point converters (ESS-Sabre.) However even those don't go past the ability of an ideal 24 bit converter.

Whether ASIO accepts 32 fixed or not is meaningless if the actual converter chip does not support it.
Old 13th January 2017
  #81
Lives for gear
 

Quote:
Originally Posted by craigb View Post
There are two things going on here, IMO. The truncation distortion itself, and then the modulation of the audio by that same distortion. That's why I made the reference to analog intermod. distortion earlier.
Distortion does not modulate. Modulation in a nonlinear process can create distortion though.

In analog, intermodulation distortion comes from one wave (or signal) modulating a second wave (or signal). The distortion is the result, not the action itself.
Old 13th January 2017
  #82
Lives for gear
 
JayTee4303's Avatar
Quote:
Originally Posted by Blaine Misner View Post
Dude. just mult the signal after the avalon. you're making a needless round trip through the convertor.
Nothing against multing, and I may go that way, but not needless.

The gain stage in the converter offers compression options for the 1176, which combines the threshold control w input gain.
Old 13th January 2017
  #83
Lives for gear
 
Brian Campbell's Avatar
 

FWIW Bitter displays 32bit output on the master fader in Pro Tools 10 (native).

I guess that it's not dithered.
Old 13th January 2017
  #84
Gear Maniac
 
Yuri Korzunov's Avatar
 

Quote:
Originally Posted by craigb View Post
Yes, I understand intermod in the analog world, consider this. You have your audio which you then truncate. This is a totally non-musical process which creates new non-musical byproducts that are added to the audio. Will the audio not be modulated by those byproducts? I readily admit that I may be thinking about it wrong, and am happy to learn.
1. Original musical signal generate multiple frequency distortion harmonics for each of original harmonics.

2. Original musical signal generate intermodulation products for all combinations of original harmonics.

Example:

Original harmonics frequencies: O1, O2.

Multiple distortion harmonics: O1*2, O1*3, ...,O1*N, O2*2, O2*3, ...,O2*N

Multiple intermodulation harmonics: O1-O2, O1+O2, ...,O1*N+O2*M, O1*N-O2*M

Intermodulations depend on non-linearity input/output-voltage curve (I don't know English term exactly).
Old 13th January 2017
  #85
Lives for gear
 

Quote:
Originally Posted by Yuri Korzunov View Post
Intermodulations depend on non-linearity input/output-voltage curve (I don't know English term exactly).
I was going to say Transfer Function but that is a term is more strictly applicable for a linear time invariant system I've learned. The correct term for a nonlinear system escapes me for the moment.

I use to be sloppy and say transfer function generally for devices or processes that accepts an input and delivers an output.

But input/output voltage curve is clear enough me thinks! :-)
Old 13th January 2017
  #86
Lives for gear
 

Quote:
Originally Posted by craigb View Post
Yes, I understand intermod in the analog world, consider this. You have your audio which you then truncate. This is a totally non-musical process which creates new non-musical byproducts that are added to the audio. Will the audio not be modulated by those byproducts? I readily admit that I may be thinking about it wrong, and am happy to learn.
I'm not sure (but someone may disagree :-) that modulation is a good term in this case. But if, I'd like to see the truncation itself as the modulation, that leads to a distorted signal.

Distortion is normally best viewed as a product of a process IMO.

If you hear an effect on the signal you hear the distortion.
Old 13th January 2017
  #87
Gear Maniac
 
Yuri Korzunov's Avatar
 

Quote:
Originally Posted by Audiop View Post
I was going to say Transfer Function but that is a term is more strictly applicable for a linear time invariant system I've learned. The correct term for a nonlinear system escapes me for the moment.
I use to be sloppy and say transfer function generally for devices or processes that accepts an input and delivers an output.
Yes. I meant: input voltage at X axis, output voltage at Y axis.

In Russian it is called as «amplitude feature/characteristic/response» (амплитудная характеристика).

But in English «amplitude response» is «amplitude-frequency response».

I just found terms in the physical and electronic dictionaries «input-output characteristic», «gain-transfer characteristic», «amplitude characteristic».

I suppose it is right terms.
Old 13th January 2017
  #88
Lives for gear
 
lectric's Avatar
 

Excelent post¡¡
Today I check the dither resolution from my Reaper 5.3 with Bitter. Generally I record at 24bits...and the internal resolution from Daw is 64 bits floats.
I see:
Master Channel (without any plugin): 64 bits
Track channel (without any plugin): before the fader 24 bits
Track channel with plugins: 32 bits or more depends of plugin..(Fabfilter Pro-Q2 32bits, SSL Drumchannel 48 bits)
Bus channel: 64 bits
Glue a track: 24 bits
Render/freeze a track: 32 bits


Conclusion:
-If I place the reainsert (external hardware out-in) after any plugin I must dither the signal to 24 bits.
-I must add a 24 bits dither immediately before the master fader for mix with the most clean a accurate sound from my Mytek 8x192 ad/da.

It´s correct????
Old 13th January 2017
  #89
Motown legend
 
Bob Olhsson's Avatar
 

The truncation also removes low level detail that is audible through the dither. If the audio has ever been truncated, the differences become much more subtle without that low level detail and the masking from the distortion.
Old 14th January 2017
  #90
Without criticizing any participant, I'd like to throw a "contrasting" article into the turmoil:

Does Dither Really Matter?

My experience definitely matches most of the experiment's results. To be clear, I'd really prefer that all this dither stuff would be abstracted away from the user, down to what it really is: A part of the sampling theorem, a technical detail. Let the operator mess with music, not PCM.

We're definitely moving at the edge of audibility, making it easy for our favorite friends to take over control of perception: https://en.wikipedia.org/wiki/List_of_cognitive_biases

In fact, the scale of dither effects (to 24bit) is comparable to the effect shapes and materials have on audio cables. Radically out of scale IMHO. To 16bit, it's certainly more relevant, no doubt about that. For 8bit it's super obvious of course. But the general nature and minuscule scale, combined with a sweet PR wording definitely provokes strong hallucinations. Let's make sure to remain aware of the problem.

In that sense, let's run a dither ABX test! (I expect "surprise")

Last edited by FabienTDR; 14th January 2017 at 01:17 AM..
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump