to dither or not to dither

[univox1]

is it a good idea to use an added dithering plug in when bounceing. such as the dither fuction on waves L-1. or the digi dithering plug in. or is it better to bounce with out?

[waterboy]

only use dither when you are going from a greater to a lesser wordlength.

For example: if you are bouncing a song recorded at 24bits to a 24bit stereo file don't use a dither plug-in. If you are bouncing a 24bit song to a 16 bit stereo file DO use dither as the last plug-in on the master fader.

I would recommend the PowR dither Digi supplies on the #3 setting (sorry I don't remember the exact term, but its the third setting on the pop-up menu in the plug)

Also, if you are sending your mixes out to a mastering lab its best to send them an undithered 24bit bounce.

Hope this helps.

[AE]

Sorry, Waterboy, your advice isn't entirely accurate.

You are correct in saying that you dither when going from a greater to lesser word length. However, you actually end up going from a greater to lesser word length most of the time when bouncing, even from 24 to 24 bits. Here's why:

If you're doing a bounce that's a mix of multiple 24-bit signals, the accumulated word length of the mix will contain more than 24 bits. If you don't use dither, you'll truncate the additional information. Of course, the low level quantization distortion and loss of dynamic range that results may or may not be audible -- and the Digi User's Guide says that dithering in this situation is not necessary -- but why, then, do most of the best-sounding plug-ins dither or noise-shape their 48-bit outputs to 24 bits? That would also apply to 24-bit bounces of mixes sent to mastering labs; once you've truncated the data below 24 bits, nothing your mastering engineer will do can restore this information. It is perfectly reasonable to dither your bounce and have your mastering engineer dither again when going down to 16 bits.

If you're doing a bounce to marry together consecutive regions into a single file, but you're not mixing, then you shouldn't dither. The only place dither should be applied in this situation is in the crossfades between regions, which is an option selected in the fade window.

[waterboy]

Hi AE,

I'll meet you halfway and say that my advice was not complete, but given the scope of the question I truncated the scope of my answer. [img]images/icons/wink.gif[/img]

[waterboy]

Hi univox1,

Dithering from 48bit to 24bit does occur behind the scenes automatically on a fair number of plugs (the Waves' double-precision plugs, the Sony Oxford, etc.)

Digi's optionally-installed dithering mixer plug-in does this as well.

Neither one of these areas offers any user control though (other than not using them at all)

As AE pointed out dither is an option when performing a crossfade and for audiosuite processing. In the Preferences menu under the Processing header you can set the audiosuite dither to match the wordlength of your session file.

[AE]

Originally posted by waterboy:
Hi AE,

I'll meet you halfway and say that my advice was not complete, but given the scope of the question I truncated the scope of my answer. [img]images/icons/wink.gif[/img]

<font size="2" face="Verdana, Arial">But I heard the truncation! Guess you weren't dithering around...

Yeah, the fact is that in many scenarios, the result of truncating whatever's below the 24th bit is inaudible to most people. But I also think that in some situations, it is clearly audible to those who listen carefully. I'd love to see someone do a real study about what we can and can't hear below the 24th bit of resolution...

[AE]

Originally posted by waterboy:
Hi AE,

I'll meet you halfway and say that my advice was not complete, but given the scope of the question I truncated the scope of my answer. [img]images/icons/wink.gif[/img]

<font size="2" face="Verdana, Arial">But I heard the truncation! Guess you weren't dithering around...

Yeah, the fact is that in many scenarios, the result of truncating whatever's below the 24th bit is inaudible to most people. But I also think that in some situations, it is clearly audible to those who listen carefully. I'd love to see someone do a real study about what we can and can't hear below the 24th bit of resolution...

[univox1]

i was mostly wondering if pro tools dithers automaticly when u bounce . and does a plug in do it better?

[Robin Hughes]

I think when you start to add many channels of dither together you may get an audible effect. After all this is a reason to use mutes at the final singnal point when mixing to stereo in analogue domain , why shoudn't this be the case with digital. I for one have been very lazy about this assuming it didn't matter. But recently , after my desk blowing a set of capicitors ( which I am sure was from a power transformer on the same ring main ... stepping down for american voltage ) , I was reduced to using internal mixing in protools as opposed to splitting out multichannel through the desk. The result from the four days work , whilst the desk was being repaired , had a crunched up quality with a lack of air around individual instruments ... even when listenned through a stereo channnel of the reinstated desk. Splitting the channels corrected the problem immedialtely.

I use the 24bit dithered mixer extension , and I realy did begin to wonder about the effect of this on internal submixing.

[tld]

Just to add to waterboy's comments on Waved plugins...

I've been using the L2 but like using the Digi POW-R for my final 16 bit dither. With a setup such as that you should (as the L2 manual states) use the L2 IDR set to 24 bit dither (to dither from the internal 48 bit to 24). I then use the POW-R at 16 bit as the last thing on the Master fader.

As waterboy pointed out, other Waves plugs that operate internally at 48 bit are automatically dithered to 24 when returning to the Mix buss.

By the way, I've been REAL happy with my finished results using the L2 and POW-R this way.

Tom