|
|
Post by kcatthedog on May 19, 2015 10:23:46 GMT -6
|
|
|
|
Post by Johnkenn on May 19, 2015 10:54:27 GMT -6
Is there really a difference in dither quality? Just wondering.
|
|
|
|
Post by warrenfirehouse on May 19, 2015 11:03:21 GMT -6
Ive read many threads and articles on dither and I honestly still have no idea what the hell it is, what it does, and why I need it. I just put it on at the end and dont think about it. Can you really hear a "better" dither?
|
|
|
|
Post by jdc on May 19, 2015 12:39:55 GMT -6
i'm by no means an expert, but it would stand to reason that you would hear differences in different types of dither, considering you're moving noise to varying parts of the frequency spectrum, depending on the type of dither being used. whether or not one is better than another would be personal preference. that being said, I was coming here to create this exact same thread haha
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 19, 2015 21:14:54 GMT -6
Since last year i follow Bob O.'s advice.
I won't go back to any noiseshaping fancy anymore. TPDF all the way long now, also on final dither.
If i look at the graphs i remember why. ~ -132dB flat. Nothing remotely harsh, strange or funny in the top end. Works for me. Period.
With the others i always had to cross check and sometimes one sounded better, sometimes another one....but no. I wanted them to sound better. They didn't...
|
|
|
|
Post by wiz on May 19, 2015 21:50:54 GMT -6
Since last year i follow Bob O.'s advice. I won't go back to any noiseshaping fancy anymore. TPDF all the way long now, also on final dither. If i look at the graphs i remember why. ~ -132dB flat. Nothing remotely harsh, strange or funny in the top end. Works for me. Period. With the others i always had to cross check and sometimes one sounded better, sometimes another one....but no. I wanted them to sound better. They didn't... can you explain, how you implement this, and what you use? cheers Wiz |
|
|
|
Post by levon on May 19, 2015 23:24:20 GMT -6
Since last year i follow Bob O.'s advice. I won't go back to any noiseshaping fancy anymore. TPDF all the way long now, also on final dither. If i look at the graphs i remember why. ~ -132dB flat. Nothing remotely harsh, strange or funny in the top end. Works for me. Period. With the others i always had to cross check and sometimes one sounded better, sometimes another one....but no. I wanted them to sound better. They didn't... can you explain, how you implement this, and what you use? cheers Wiz I second this request  |
|
|
|
Post by swurveman on May 20, 2015 6:50:01 GMT -6
can you explain, how you implement this, and what you use? cheers Wiz I second this request Hell, I third it. |
|
|
|
Post by goodhertz on May 20, 2015 10:27:50 GMT -6
I second this request Hell, I third it. Hi all, Good Dither has an option for Flat TPDF Dither (just set the noise shaping to “none”). Cheers, Devin Kerr Goodhertz |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 20, 2015 11:07:39 GMT -6
The implementation is as easy as it gets.
Sonar has an option to apply dither that can be set and forget, in the meantime they set it to TPDF as default. In the export dialogue, it can be set explicitly, too.
Harrison Mixbus shows the option to apply the dither of your choice in export dialogue, too.
Other DAWs should have a preferences setting for the dither as well.
Since any modern DAW should use TPDF on any occasion dither is recommended and necessary anyway up to the export, most DAWs give only the option to change the final export dither (or none).
For export, if your DAW supports additional noise shape dithers like the free ones, POW-R, MBit or whatever, or no dither, in case you use the dither option of a mastering plugin, they give you the option to choose explicitly.
If i go the full mile for mastering a song in an explicit mastering session, i export with TPDF and nothing else applied out of the DAW anyhow, because this is, what usually is recommended by any mastering engineer AFAIK, because he will do further dither demanding processing, too, and apply the final dither, which is the only one, where different dither could have made sense.
For these occasions i use an external command line program for processes like bit reduction (and it's necessry dither application) and SRC, like finalcd and SoX, which can also be used for batch processing.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 20, 2015 11:35:37 GMT -6
Edited after Bob O.'s comment to correct/clarify.
So, an ideal DAW should apply TPDF dither before any bit depth changing file export or DA automatically.
Setting dither with a plugin:
Audio -> mastering plugin processing with final dither application -> export without additional dither
AFAIK most DAWs do that automatical dithering at any volume change to file or DAC or whatever process where dither is needed. But - they don't tell you details about it. So i don't know if they are all doing it right. They should, at least.
As of Bob O., this seems NOT to be the case with most DAWs. Which is...odd.
Please correct, if i something is unclear or plain wrong, if you know better. We have quite some DSP cracks on the board, that might know better, last but not least Bob O. ....
|
|
|
|
Post by Bob Olhsson on May 20, 2015 12:07:18 GMT -6
You don't want to d*ther before processing, just after processing before bit reduction to a file or D to A so it won't distort. Doing it before just adds unnecessary noise. Unfortunately Samplitude is the only DAW that has smart d*ther I'm aware of.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on May 20, 2015 13:01:35 GMT -6
Thank you, Bob, for your input. Clarifies it for me.
As always.
Very appreciated!
|
|
|
|
Post by Johnkenn on May 20, 2015 13:06:31 GMT -6
So - I'm a little confused. When I bounce my mix down from 24/whatever to 16/44.1, are you saying just put dither on the master track in the last position (like I've always done) or put dither on every track?
|
|
|
|
Post by Bob Olhsson on May 20, 2015 13:07:36 GMT -6
It was all explained to me by Jim Johnston of Bell Labs during the '80s. Unfortunately very little accurate information ever made it into manuals or magazines.
|
|
|
|
Post by jdc on May 20, 2015 13:17:58 GMT -6
Hell, I third it. Hi all, Good Dither has an option for Flat TPDF Dither (just set the noise shaping to “none”). Cheers, Devin Kerr Goodhertz David, welcome to the board! I've been loving your plug ins lately, well done sir. |
|
|
|
Post by Bob Olhsson on May 20, 2015 13:22:20 GMT -6
You only need d*ther on something being written to a file or sent directly to a D to A. We need to trust plug-ins handling their internal bit reductions properly. Old Pro Tools TDM-HD took care of d*thering to plug-ins with a dithered mixer plug-in. Native DAWs run at some combination of 32-80 bit float which requires d*ther to 24 bits or 16 depending on what you're doing.
|
|
|
|
Post by watchtower on May 20, 2015 14:25:26 GMT -6
You dither the final render, not the individual tracks. Everything goes through the master buss at 32-bit float or higher. Then you dither down the master output to 24 or 16 or what have you. I use noise shaping dither and never noticed any issues. It's not a huge deal, honestly. I bet there are many times when you wouldn't even be able to hear a dithered final file, versus a truncated one on a loud rock mix, but in theory, you should always dither when reducing the bit depth to alleviate quantization error through randomization (noise, AKA dither).
EDIT: As far as I know, ALL converters have a natural noise floor that is actually above the 24-bit noise floor. You're never truly hearing 24-bits worth of audio. There is always going to be an inherent level of noise/dither in the DAC due to the limits of analog. I think it's debatable whether dithering 32-bit floating point to 24-bit fixed point is even necessary. And when you listen through your DAW using 24-bit converters, you're always going to have that natural level of dither applied by your DA. I don't think it's necessary to dither your Master bus while mixing, but it is ideal when rendering down to a 16-bit file. You could probably get by by only dithering when you go down to 16-bit
|
|
|
|
Post by Bob Olhsson on May 20, 2015 14:45:55 GMT -6
I think d*ther is more important today than it was during the '90s because today most audio is played with dsp volume and tone controls. Every time the signal gets truncated without d*ther, the crunchiness builds up by around 6 dB. Where you won't immediately hear that much benefit is vis which are truncated all over the place.
Truncation destroys the depth in loud rock mixes.
|
|
|
|
Post by Pueblo Audio on May 20, 2015 17:14:36 GMT -6
Ive read many threads and articles on dither and I honestly still have no idea what the hell it is, what it does, and why I need it. Most people associate dither with word length reduction. While dither is part of that process, there is a more global purpose for it: to alleviate Quantization Distortion. When an analog signal is sampled, the amplitude of each sample is measured and assigned a quantity, in bits. But if that signal's amplitude falls between two fixed quantity values, then the signal cannot be represented accurately in the digital domain. The digitized signal is therefore distorted. This is Quantization Distortion. By applying specific types of low level noises, QD can be reduced or even eliminated. Thermal noise floors of analog equipment is generally NOT effective in eliminating QD. So anytime a digital signal is processed and a new amplitude values is to be assigned, there will be some resulting amount of QD. So some amount of dither may be applied to relieve that distortion. Many digital processes are actually applying dither in the background, all the time. Happily, in this age of floating point systems, that QD is pushed way, way down. As long as a signal remains in a floating point environment, dither may be delayed until exported back to a Fixed Point state and/or reconstructed back to an analog signal. The more interesting question/topic might be; "What does quantization distortion sound like?" |
|
|
|
Post by Bob Olhsson on May 20, 2015 17:16:38 GMT -6
What the d*ther does is randomize the bottom bit when it gets lopped off so it doesn't chatter along with the signal like a noise gate. Modern over-sampling A to Ds d*ther to 24 bits so there's no need to worry about that. There's no point in recording an A to D converter at more than 24 bits. It's when you start writing DSP calculations to a file that 32 float becomes useful. My tests with post 2000 converters have shown TPDF to retain detail better. Noise shaping could sound better with earlier converters but at this point I'm not convinced it wasn't just a pleasant sounding band-aid masking some other problems. I mean after all it is way down there in level if, and it's a great big if, everything is working right. We now know a lot of digital audio stuff wasn't working right during the '90s.
|
|
|
|
Post by Bob Olhsson on May 20, 2015 17:26:15 GMT -6
By the way, it's just lazy programming that we even need to think about d*ther. It would be like having to always remember to turn the bias on when putting an analog recorder into record.
|
|
|
|
Post by wiz on May 20, 2015 17:32:07 GMT -6
So, let me ask basic questions to make sure i understand this.
With a mix, in a Modern DAW (e.g. Logic Pro X, Cubase (whatever the recent version is etc)
1. You are exporting a mix for mastering at the same bit depth and sample rate as the project (e.g. 96Kz 24bit)
and that is what the mastering house is using, you should NOT apply dither?
2. You only need to apply dither, in a Modern DAW during the MIX process, if you are doing a D-A , e.g. using a I/O plug in or routing out and
back into you Converter to apply some sort of outboard processing?
2a) if question 2 is true.. and I need to dither going out for re amping using hardware, how do I achieve that? I would have to apply some sort of plug in, that has dither, prior to the I/O plug in? if so what plug in do you recommend to do that?
3. You don't need to put dither plug ins , within your plug in chain on a track within the DAW during mixing, you have to trust each plug in does any
dithering within itself for you?
4. If your mix session was 32 bit, and you are exporting to 24bit for mastering do you apply dither?
-----------------------------------------------------------------------------------------------------------------------------------
There is often a lot of technical discussion around dither.. but not much about how to exactly implement in within the mix you are doing today to export to the Mastering house, this afternoon... and exactly how to achieve that with the current technology.
I have never implemented manual dither of something during the mixing process ever.
I only apply dither here, when making CDs in house. Other than that, things go to mastering at the bit depth, and sample rate of the project and I leave that up to the mastering engineer, all my projects for as long as I can remember have been done at 24 bit.
cheers
Wiz
|
|
|
|
Post by Pueblo Audio on May 20, 2015 18:20:01 GMT -6
Whether to dither or not is a question of one's perception and sensitivity to quantization distortion vs. noise. I don't believe there is a global answer. One needs to respond to what they hear before them and decide for each specific situation. Unless you spend time learning to identify this distortion, it will be hard to answer that question for yourself.
As far as delivery to mastering: if not confident about dither choices, it is always possible to export mixes as 32f files for delivery. Then you can let the mastering house sort it out. This will depend on how your mastering house works, so simply ask them.
|
|
|
|
Post by tonycamphd on May 20, 2015 18:20:37 GMT -6
So, let me ask basic questions to make sure i understand this. With a mix, in a Modern DAW (e.g. Logic Pro X, Cubase (whatever the recent version is etc) 1. You are exporting a mix for mastering at the same bit depth and sample rate as the project (e.g. 96Kz 24bit) and that is what the mastering house is using, you should NOT apply dither? 2. You only need to apply dither, in a Modern DAW during the MIX process, if you are doing a D-A , e.g. using a I/O plug in or routing out and back into you Converter to apply some sort of outboard processing? I understand that if you have made any tweaks ITB prior to DAing, you are subject to the QE's, and SHOULD apply a dither to all channels that are tweaked heading to DA, my concern is that using a dither plugin instantiated on 32 DA's into a hybrid console set up, will allow the low level dither noise to stack and build up to audible levels... is this the case? I've been unsure of this for some time.
2a) if question 2 is true.. and I need to dither going out for re amping using hardware, how do I achieve that? I would have to apply some sort of plug in, that has dither, prior to the I/O plug in? if so what plug in do you recommend to do that? 3. You don't need to put dither plug ins , within your plug in chain on a track within the DAW during mixing, you have to trust each plug in does any dithering within itself for you? 4. If your mix session was 32 bit, and you are exporting to 24bit for mastering do you apply dither? ----------------------------------------------------------------------------------------------------------------------------------- There is often a lot of technical discussion around dither.. but not much about how to exactly implement in within the mix you are doing today to export to the Mastering house, this afternoon... and exactly how to achieve that with the current technology. I have never implemented manual dither of something during the mixing process ever. I only apply dither here, when making CDs in house. Other than that, things go to mastering at the bit depth, and sample rate of the project and I leave that up to the mastering engineer, all my projects for as long as I can remember have been done at 24 bit. cheers Wiz good post wiz.. |
|
