Question on dither

[gubu]

Hi forum,

A quick question on dither.

 

When bouncing down a 24 bit stereo file from a 24 bit recording session, do you need to apply 24 bit dither or does this matter? I've been doing a little googling and searching HC forums but it's unclear whether you need to apply dither when bouncing to the same wordlength.

 

I'm working in ProTools LE 7.4.

 

Thanks in advance for your replies,

ad

[WRGKMC]

What many do like myself, is mix the song down at the same recording sample rates.

You then master it at those higher sample rates to maintain the highest quality possible.

When you're done mastering you then convert the sound to 16/44.1 and apply the dithering

at the same time as the very last step when all processing is done.

 

If you dont master your music or master while mixing and dont intend to do any other processing

then you can go ahead and apply it mixing down.

[gubu]

That's exactly what I do. I record/mix in 44.1k 24bit, bounce to stereo in 44.1k 24bit and apply 16bit dither when bouncing to 16bit audio file after any mastering processing. (I use 44.1k as my old Mac doesn't have the horsepower to run mixes with plugins at any higher sample rate and I don't trust the ProTools LE system as an accurate SRC.)

 

My question is do I need to apply 24bit dither when bouncing a a 24bit mix to a 24bit stereo file?

 

We were taught in college that dither was unnecessary when bouncing to the same bit depth/wordlength. I've been doing a lot of research lately that suggests that quite a lot of what was taught on the (fully certified!) sound engineering courses I've done is, at best, unreliable and inaccurate, and at worst, plain wrong.

Hence, the noob nature of my question! I'm getting back to first principles here to make sure I can get the most sample accurate and distortion-free workflow going that I can...

I've found a lot of great information by searching the HC forums and elsewhere and also from a couple of oft-recommended pro-audio 'bibles' but haven't been able to find anything concrete on this question...

[Zooey]

 

My question is do I need to apply 24bit dither when bouncing a a 24bit mix to a 24bit stereo file?

 

 

You don't need dither until you go from 24 to 16 bits. Keep your "premaster" bounce in 24 bit for as long as you can--any mastering type processing should be done on the 24 bit file. Only apply dither at the very last step of preparing to burn to CD.

 

This raises a good question: if the delivery format you're working on is MP3 (like a non-final bounce for listening to on your iPod), is dither recommended? I honestly don't know. There are plenty of people who rarely dither, by the way. Some people whose opinions I respect think that under certain circumstances, a truncated mix sounds better. Believe it or not. I just dither all the time when going to 16 bit to be safe.

[gubu]

 

You don't need dither until you go from 24 to 16 bits. Keep your "premaster" bounce in 24 bit for as long as you can--any mastering type processing should be done on the 24 bit file. Only apply dither at the very last step of preparing to burn to CD. This raises a good question: if the delivery format you're working on is MP3 (like a non-final bounce for listening to on your iPod), is dither recommended? I honestly don't know. There are plenty of people who rarely dither, by the way. Some people whose opinions I respect think that under certain circumstances, a truncated mix sounds better. Believe it or not. I just dither all the time when going to 16 bit to be safe.

 

 

Yep, I'm good with all of that. 16 bit dither should be applied once and once only when all mastering processing has been completed at 24 bits.

 

Here is the thing tho:- ProTools LE is using 32bit floating point on the mix buss and AFAIK some RTAS plugins also use 32 bit floating point. So, should we apply 24 bit dither when bouncing a 24bit mix to a 24 bit stereo premaster to avoid truncation?

I know 32 bit floating is different from 32 bit fixed but it seems to me that 24 bit dither at this stage may be sensible if you want to avoid truncation?

[Zooey]

 

Here is the thing tho:- ProTools LE is using 32bit floating point on the mix buss and AFAIK some RTAS plugins also use 32 bit floating point. So, should we apply 24 bit dither when bouncing a 24bit mix to a 24 bit stereo premaster to avoid truncation? I know 32 bit floating is different from 32 bit fixed but it seems to me that 24 bit dither at this stage may be sensible if you want to avoid truncation?

 

 

PTLE uses 32 bit floating point internal architecture for processing, but the tracks always stay in 24 bit and the result of the bounce is 24 bit. Nothing is being truncated at the bounce stage.

 

I've heard this argument before, but it seems really esoteric and not worth worrying about. I don't even know if PTLE has a 24 bit dither option. Does it? Most dither algorithms go up to 16 or 20.

[gubu]

PTLE uses 32 bit floating point internal architecture for processing, but the tracks always stay in 24 bit and the result of the bounce is 24 bit. Nothing is being truncated at the bounce stage. I've heard this argument before, but it seems really esoteric and not worth worrying about. I don't even know if PTLE has a 24 bit dither option. Does it? Most dither algorithms go up to 16 or 20.

 

So that should mean that there is some sort of internal dithering going on between the floating point algorithms and the 24 bit output? If so, then I'd have to agree about it not being worth worrying about.

 

Yes, there are a few RTAS plugs that give an option of 24bit dither. Waves L2, LinMB and Pow-r Dither are some that spring to mind...

I'm wondering now tho why Waves would offer 24bit dither for use with equipment that can't operate at any higher bit depth. Maybe I should go and read the manual

[LeonardScaper]

I rarely dither these days. Everything is going out as MP3 and there's already plenty of ambient noise in my stuff.

 

When I do an album (CD) I dither with Pow-R Dither. Type 1? Type 2? Who knows?

[WRGKMC]

Yes waves does have some dithering options in they're plugins. I've always wondered about them myself. I cant hear a difference between using them or not. Maybe if they were used in a daw as mastering plugins and you export to 16 bit they have some effect, I dont know. I do all my mastering in a separate editing program so its no issue for me.

 

The quality of the dithering program is supposed to have a big impact on the sound quality.

I been using cool edit for an editor program and it has several different types of dither.

Sonar also has several including power 1,2 & 3. I can tell theres some very mild differences using those in the high frequency cymbals and such, but it is real mild and not obvious on all material. To me its like rendering a photograph after editing to smooth the edges and hide the editing that was done. Its pretty much the same deal with audio.

[Zooey]

 

So that should mean that there is some sort of internal dithering going on between the floating point algorithms and the 24 bit output? If so, then I'd have to agree about it not being worth worrying about.

 

 

Due to the fixed point math of the TDM bus, the PTHD dithered mixer actually does insert dither at certain points of the signal flow. Apparently some hardware digital mixers do it, too. PTLE does not. It's been 10 years since this topic was hotly debated, but my recollection is that internal calculations in 32 bit float doesn't require it.

[Mo Facta]

It's been 10 years since this topic was hotly debated, but my recollection is that internal calculations in 32 bit float doesn't require it.

 

This is because in 32-bit floating point audio the noise floor is proportionate to the signal. Thus, we have a shifting noise floor which does not require dither. It is also helpful to note that all audio within your session is always 24 bit regardless of the internal bit depth of the session.

 

Dithering shouldn't have any effect on frequency content (directly, anyway) but instead on the resolution of a low level signal. This means that without dither the quietest of sources that do not reach the first quantization tier will manifest as a square wave and hence truncation distortion. Dither is nothing more than randomized noise that is introduced to the noise floor to smooth the attenuation of the signal beyond the actual limit of the lower bits with the added trade-off of a higher noise floor.

 

I would say that noise shaping algorithms should have more of an impact on the frequency domain as it is undetermined how drastically the shifting of this noise to the upper registers affects the aesthetics of the resulting audio.

 

I would also say forget about dither and just get on with it. If it sounds good, use it. If it doesn't, don't.

 

Cheers

[Johnny-Boy]

Dither is always an interesting topic to me. Maybe because I can't here any difference when apllying dither, while others can.

 

Here are two short identical tracks. One has dither added, the other doesn't. All observations will be appreciated.

 

Best, John:cool:

 

Test 1

http://schicksville.com/Music/T-1.mp3

 

Test 2

http://schicksville.com/Music/T-2.mp3

[Mo Facta]

The way I had dither demonstrated to me was with the Prism Sound DScope software by Graham Boswell.

 

What he did was he created a 16-bit environment in DScope and turned the volume right up so that we could hear the absolute digital noise floor. This inherent noise is actually the sound of the lower bits in digital "silence" being truncated around -93dBfs (-96dBfs theoretical) - the lower limit of dynamic range in 16-bit. You can hear the same thing if you render a 16-bit file of digital silence and turn it up loud (provided the noise floor of your analog components do not exceed it).

 

He then introduced a VERY quiet 16-bit test tone of 1kHz to the noise (around -85dBfs or so) and attenuated it so that it's level diminished below the -93dBfs noise floor. Without dither the tone completely squarewaved at it crossed the noise threshold because, as I said above, the first quantization tier was not reached and the signal manifested as truncation distortion. This is the real effect of not dithering and why it is only the extremely low level stuff that gets affected, like reverb tails and cymbal decays. So, if you're producing really sparse music with many quiet passages I would say that dither might be more important to you.

 

After that, he introduced dither (the Prism Sound variety, of course) and we could clearly hear the increased noise floor, which now jumped to about -90dBfs. He once again introduced the quiet 1kHz tone and attenuated it until it reached the noise and low and behold, it diminished smoothly and with linearity below the noise into nothingness without any artifacts or distortion whatsoever. He then introduced noise shaping to the signal which pushed the excess noise to the upper register and virtually disposed of it. What we were left with was a clean, clear and smoothly diminishing 1kHz tone in an environment that actually EXCEEDED the noise limits of 16-bit audio. This is the reason why many dithering and noise shaping systems claim that you can achieve 16-bit audio that actually sounds more like 18-bits.

 

His conclusion was that with the proper dithering and noise shaping considerations, we have ALL the resolution we need in 16-bit audio. After it was demonstrated to me, I am inclined to agree with him.

 

However, just how all this affects a mix or music aesthetically is anyone's guess. That's up to you.

 

Cheers