Audition 3.0 - 16 or 32-bit?

[Guest]

Today at 07:19:11 AM

Hi All

A novice here - I’m using Adobe Audition 3.0 to produce various things for a radio station that I volunteer at.

My source files are WAV (44100Hz, 16-bit). The finished mixes also need to be the same due to the playout system that the station uses.

There’s an option in the settings to “auto-convert all data to 32-bit upon opening”. Is there any real benefit in enabling this? Will doing this improve how Audition applies effects (like reverb, compression etc.) or will it make no difference? My understanding is that AA's audio engine works in 32-bit float anyway (regardless of what file you feed it) so there's no point.

When create a mix down, AA converts it to 32-bit. Is it worth just disabling that so they come out as 16-bit? My understanding is AA's audio engine when adding effects etc works in 32-float anyway in multitrack and edit view?

Also, what I tend to do with effects is apply them in the edit view first as I find it easier than using the effect rack in the multitrack area. Is there any benefit to applying them in the multitrack view instead...in terms of quality? The reason I ask is I believe multitrack does 32-bit mixing (at least it says that at the bottom of the screen).

Last question – I read this online;

Quote

Dither Transform Results (increases dynamic range) - disable it
This feature is described like a good thing...really it means that anytime you make any editing there is a dithering applied, for example you edit/process only a small segment and you have that segment with dither too... or you edit/process something twice or more times and you have dither applied twice or more times too. Very nasty idea!

[SteveG_AudioMasters_ · Accepted solution]

The major benefit of mixing in 32-bit floating point is that you have got away from the major limitation of 16-bit integer files - let's just look at that for a moment. If, for instance, you reduce the level of them and save the result, you will be reducing the bit depth available - permanently - by one bit for every 6.02dB you reduce it. And inevitably, you'll be raising the noise floor by this amount too. Possibly not too bad on an individual file, but if you have several of these in a mix, then the background noise floor will be significantly raised - dB addition of noise is cumulative.

But when you operate with 32-bit floating point files, this problem is eliminated. The files are scaled, and have an effective dynamic range that's so great (into four figures) that it's not worth even considering. And if you alter this scaling (by running them at different volume levels), it doesn't alter the fundamental signal data at all, so any amplitude change you make can be undone without loss. And this includes positive ones too - you can set a signal so that it's massively overloaded, save the result, and upon reopening you can reduce the level again and have your original signal back. No it's not magic, but sometimes it can seem like it!

Operating the mixdown as 32-bit FP does have some benefit when effects like reverb are added - the reverb tails aren't lost, whatever you do, for instance. As for your quote - well it's definitely true if you are applying effects to a 16-bit file, but with a 32-bit floating point mix you won't get dither applied anyway. Dither only becomes an issue when you save your final result as a 16-bit integer file again, and the important thing about it is that you should only apply it once, otherwise you end up with an inevitable increase in the noise floor. So in principle, it's much better to apply effects in a 32-bit mix situation, get everything the way you want it in this format, mix it down in this format, and then use 'save as' to store a 16-bit integer copy, where dither is applied to everything - just the once.

Parts of it at the start are slightly misleading (try to ignore the  'join the dots' bit as it doesn't work quite the way they've described it, although the outcome is correct) but an otherwise good guide to what's going on with dithering was produced by iZotope. Whilst it's primarily about Ozone, the principles apply to all audio. You can download the pdf of it here: http://downloads.izotope.com/guides/izotope-dithering-with-ozone.pdf

[Guest]

Hi Steve

Thanks for the detailed reply.

This is how I plan to work;

Convert all my master source files to 32-bit. Do my edits / multitrack stuff, mix it down to one file, then convert the finished mix down to 16-bit.

That way, I'll still have my original 32-bit masters and I'll only be dithering once.

Is that a good way to work?

[SteveG_AudioMasters_]

damient24997891  wrote

Hi Steve

Is that a good way to work?

Yes!

[Guest]

Thank you sir.

[MusicConductor]

I'll put it to you this way: it's such a great way to work that all current versions of Audition do this automatically without being asked.

AA3 is great, don't get me wrong; it's the ultimate version back in the days of Windoze XP.  But both Windows and Adobe have moved on, and hopefully in the future you'll have a chance to benefit by moving up with them.

[wavingthru22]

Would of just been a lot easier if you just gave a simplified answer instead of over explain everything.