To Dither Or Not To Dither

[807Recordings]

Ok so here is a question. So I know in mixdowns of the final stages it is good to dither before it goes to mastering. Some say don't use noise shaping but what about if you do bounce downs in a program like Logic with each individual audio track or vst instrument being bounced by themself?? Of course the rest of the parts would be mixed down in another program but should I dither or not on the individual parts? Normally I do when working at 24/96 out of logic with pwr-1.

Can one over do the dither this way and what are the drawbacks, if any??

[D.A.S.]

if you have not read it already, here is bob katz article on how to and when to dither:

Digital Domain - Dither

Hope this helps!

Cheers!

[Mike H]

It is not clear to me why you are bouncing the individual tracks.
You don't mention what the other program is.................
can't you just import the individual tracks into the program in which you are doing the mixing?

Anyway maybe this will help some:

When to Dither?

[SeanG]

Originally Posted by 807Recordings

Ok so here is a question. So I know in mixdowns of the final stages it is good to dither before it goes to mastering. Some say don't use noise shaping but what about if you do bounce downs in a program like Logic with each individual audio track or vst instrument being bounced by themself?? Of course the rest of the parts would be mixed down in another program but should I dither or not on the individual parts? Normally I do when working at 24/96 out of logic with pwr-1.

Can one over do the dither this way and what are the drawbacks, if any??

To answer you question simply, a track should only be dithered once and it should be the last thing that ever happens to the track.

So when you are sending a mix for mastering, do not dither. tutt

If you are making changes to a track that has already been dithered go back to an un-dithered version to make your changes, and then dither over as the last step.

[807Recordings]

Originally Posted by Mike H

It is not clear to me why you are bouncing the individual tracks.
You don't mention what the other program is.................
can't you just import the individual tracks into the program in which you are doing the mixing?

Anyway maybe this will help some:

When to Dither?

Usually I do final stuff in sequoia or samplitude. There are some tools I can use for phase and I just find I get consistently better mixdowns in that program. Everything stays 24-96 right to the cutting of the vinyl lacquers.

[SeanG]

Originally Posted by 807Recordings

Usually I do final stuff in sequoia or samplitude. There are some tools I can use for phase and I just find I get consistently better mixdowns in that program. Everything stays 24-96 right to the cutting of the vinyl lacquers.

If you are not reducing bit depth, than you do not need to dither.

Final mixes should not be reduced anyway, it should be done at the end of mastering, not mixing.

Furthermore, if your tracks are going to vinyl only, you should not truncate or dither.

[Strobian]

Originally Posted by Mike H

It is not clear to me why you are bouncing the individual tracks.

In logic you can not record sub mixes to a bus like Protools or Digital Performer. So if you have audio stems of drums, you would have to bounce all those individual drum tracks to a submix if you wanted to compress them together or apply effects or whatever.

+1 on Digital Domain - Welcome to Digital Domain!. Bob Katz's articles are great.

turn your dither off.

[807Recordings]

Originally Posted by Strobian

In logic you can not record sub mixes to a bus like Protools or Digital Performer. So if you have audio stems of drums, you would have to bounce all those individual drum tracks to a submix if you wanted to compress them together or apply effects or whatever.

+1 on Digital Domain - Welcome to Digital Domain!. Bob Katz's articles are great.

turn your dither off.

Bob Katz article is good. I also know that about submixing I am not totally new to this game. It would be nice though if logic included phase controls with Xover ranges. Often synths sound best in stereo but that is hell on a bass line so its nice to be able to just mono the synth from 200Hz down and leve the rest in stereo. Sometimes even tighten up a few sounds in the midrage but add some extra spread to things like ambient reverbs. Like most things it is nice to be able to do this on a per track basis and one of the major things I like in Sequoia. Of course that is not the only reason as I mentioned for mixing down in that program.

As for the record for the most part I have not dithered when at 24-96 and if so it has been done just before the final mixdown. I never go to 16-44.1 as that just kills the music (for me). Even the masters copy I get back from the house sounds smaller compared to the 24-96 masters and of course the vinyl.

[807Recordings]

So after listening to some advice I went back and did two versions of the mixdowns from logic pro.
One version I did no dither right out to the final output stage and mixdown in Samplitude. The second version I did Dither only on the master and the third was dither at logic stage on a each part. All files remained 24-96 all the way through. From what I can tell listening here the one the version that had no dither sounded the cleanest and most detailed. The worst seemed to be the one where I used dither at logics stage on each individual part. I had to listen carefully but when I used either my Shure e5's or the Accentus Ribbon speakers I could hear a slight veil over the music and perhaps a bit of haze. Things just kind of blurred ever so slightly on the high end. This seemed noticeable when I had 32 beat cymbal rolls from the 909 drum machine. Though subtle some of the sound just blurred and lost its edge.

Either way the mastering engineer will get the non dithered file.

Thanks for the suggestions.

[postfader]

The rule for dither is that if you are changing the bit depth, e.g., reducing it from 24 to 16 bits, then you need to apply dither. To complicate things further, some plugins will have a dither button on them because they are changing the bit depth (and may also be working at a different sample rate). Many people prefer to use a plugin to reduce the bit depth, e.g., from Waves, that let's them apply and listen to the different dither types during playback (instead of applying it a bounce time as Logic does). While the PowR dither algorithms are good, they are certainly not the only ones on the market.

It's hard to predict what kind of dither function will work best with different kinds of music and mixes. So often the types of dither you can switch between in a plugin are just labelled "type 1, type 2, type 3", and you need to use your ears to determine which sounds best. Noise shaping is another option you can select from among different types when changing the bit depth - and again auditioning them is prefereable together with the different dither types.

[807Recordings]

Originally Posted by postfader

The rule for dither is that if you are changing the bit depth, e.g., reducing it from 24 to 16 bits, then you need to apply dither. To complicate things further, some plugins will have a dither button on them because they are changing the bit depth (and may also be working at a different sample rate). Many people prefer to use a plugin to reduce the bit depth, e.g., from Waves, that let's them apply and listen to the different dither types during playback (instead of applying it a bounce time as Logic does). While the PowR dither algorithms are good, they are certainly not the only ones on the market.

It's hard to predict what kind of dither function will work best with different kinds of music and mixes. So often the types of dither you can switch between in a plugin are just labelled "type 1, type 2, type 3", and you need to use your ears to determine which sounds best. Noise shaping is another option you can select from among different types when changing the bit depth - and again auditioning them is prefereable together with the different dither types.

Well with my experience as I rarely go down to CD depths I found that the dithering did add a level of haze. Not much mind you but when I compared the two mixes there was something in the air that was not there. I try and use pluggins sparingly and even when using them try and choose them carefully. Much like when I use and used hardware I listen before and after. Other than Rbass I rarely use waves and most my music is straight dance floor techno though I do stuff with lots of ambience and then I find it depends what dither I use.

If I am making a CD its a whole other story.

Either way some great suggestions on here and I thank everyone for the responce. IT DOES make a better record.

[peeder]

Originally Posted by postfader

The rule for dither is that if you are changing the bit depth, e.g., reducing it from 24 to 16 bits, then you need to apply dither.

Yes, and only then.

To complicate things further, some plugins will have a dither button on them because they are changing the bit depth (and may also be working at a different sample rate). Many people prefer to use a plugin to reduce the bit depth, e.g., from Waves, that let's them apply and listen to the different dither types during playback (instead of applying it a bounce time as Logic does).

The plugins are not changing the bit depth. The bits are truncated at bounce time. The dither plugin applies dither to the 24bit (or actually 32 or 64bit floating point or 48bit integer, depending on DAW) data before it gets to the truncation stage, and if you don't truncate and leave the file at 24 bits, you will have simply added noise at the 16th bit and will get no benefit at all.

Do not dither unless you are truncating. Always dither immediately prior to truncating.

[postfader]

Originally Posted by peeder

Yes, and only then.



The plugins are not changing the bit depth. The bits are truncated at bounce time. The dither plugin applies dither to the 24bit (or actually 32 or 64bit floating point or 48bit integer, depending on DAW) data before it gets to the truncation stage, and if you don't truncate and leave the file at 24 bits, you will have simply added noise at the 16th bit and will get no benefit at all.

Do not dither unless you are truncating. Always dither immediately prior to truncating.

Truncation means that the bits which cannot be represented in the lower reolution format are just chopped off/lost, with no attempt to account for their value, which results in the worst error (before dither is applied). This technique should never be used.

Truncation is not what mathematically happens when floating point values are properly converted to lower resolution fixed point values (for say 24 bit fixed or 16 bit fixed bit depths). The proper way to convert is to round the higher resolution values up or down to the closest value at the lower resolution (and also to ensure that there is no repeating pattern in even the error produced by rounding).

More then a few plugins work at higher resolutions or bit depths then the native DAW software (or native plugins allow). Izotope's Ozone does everything at a higher resolution (64 bit float) before converting back to a lower resolution (e.g., 32 bit float), and passing the audio data on to the next plugin. Similarly, the Waves linear phase plugins have a dither button on them to give you the option to convert from their internal 48 bit format to 24 bit (if your DAW does not support 32 bit float). I'm not sure what Apogee's UV22 exactly does, but suspect it's a similarly smart conversion process.