Hi there, Im looking into the best way to export files for mix. after a while i wanted to be sure that batch export and one track at a time has the same quality. took one channel and exported it regularly through stereo out, and the 2nd export via batch and only selecting said channel. i thought that in the null test they would cancel each other being the same, but nope. there is a difference…what am i missing here? i understand one goes through stero out but should it change the sound? do i need to export one by one? thanks for your answers. hope my q is clear.
How is the difference quantified? Or what is its value?
Is your master doing anything? Any gain change? Any plugins on it?
Hi! thanks for the reply
i did a null test. pulled the 2 files to a new project and flipped the phase on one of them. if they were the same - there should be silence.
nothong on the master channel. set to 0db
I know, but what was the actual quantifiable result specifically? Did you hear something? If so what did you hear? How loud was it? Did you hear nothing at all but saw it on the meter? How low was it? And so on…
For example, dither will live really far down in your signal and it can show up after a null test simply because dither minus dither does not equal zero. But the dither noise should be very, very low, around -144dBFS for 24 bit fixed point. In this case the files “don’t null” but the difference is of no consequence and the content that you actually care about “in” the signal is fine.
Just to clarify: You used Export Audio Mixdown both times, once with Output Channel selected and once with Audio Channel selected? If I do this, the resulting files are identical to the bit.
the audible difference within the test was not subtle. high frequency and at times almost 50% of the audio was audible, and in pretty high volume. (acoustic guitar if that matters)
as im trying to send stems for mix i tried various ways, bounce selection, export etc. and each way is a bit different and i cant tell if for the best, worse, or if it matters at-all.
And do you have anything at all on your source audio tracks?
If you do, try using the same method to export both times. If the effects on the source tracks don’t track 100% every single time you play back then the outputs are going to differ even though you’re using the same method.
I’m not sure if I’m completely understanding what you are exporting in the batch process, but, at least in my experience, it would be rare for combining stems to 100% match the full mix through the stereo out.
For one thing, the mix of the stems (which I am exporting through the stereo out as I want all submixes, including master processing, to be included in the stems) ends up being significantly louder than the original mix. Why? Because mastering compressors and limiters won’t be hit by the levels of the full mix when rendering each stem, so those processors will not compress/limit as much for the single stem as they would have effectively done when that stem was just one component of the full mix.
I’ll also have a (full) mix bus in between any group/stems buses and the master bus (which is the Stereo Out in my case), and that can have similar effects, as can any group buses prior to that, depending on what is going on at each level.
While those would probably be the main contributors to adding of stems not being identical to the full mix in my case, another possibility can be effects (or even virtual instruments if they’re not rendered to audio or frozen ahead of the export operation) that have any level of randomization in them (i.e. where rendering them two times in sequence will rarely or never produce the exact same result).
If you’re trying to get things to be identical in scenarios like my common one, you’d likely need to avoid having any dynamics inserts on any group channels (including the Stereo Out) downstream of whatever is going into each stem, while also making sure that any sends you’re using (e.g. for reverb) are included at the stems level rather than downstream (e.g. going directly to a mix bus) and that nothing that isn’t rendered to audio ahead of the export has any sort of randomization. There may be ways to work around these sorts of considerations by clever use of sidechains, but I’ve never bothered experimenting with those as my summing stems (after turning the volume down) seems sufficiently close to my normal master for most purposes.
Thats the thing, Theres nothing on the channels or the master. my goal here is to find the best way to prepare stems, clean ones, no pan, no eq, no nothing to send for mix. let the mix guy do the work from zero. but i keep getting slightly different wavs in every method i use.
a real head scratcher,
thanks for the reply,
im keeping it way simple.
no groups, no plugins and no eq. nothing on the master. trying to send clean stems for mix (stems in my case - channels. all audio)
trying to bounce all of them and grabbing them from the audio folder. but when testing them before the bounce and after - theres a difference as mentioned. trying to export batch - difference between original files and exported ones.
im really lost with this. i think that eventually it dosent really matter and every way would work, but i hate the thought of sending channels of less quality.
Could you please just post images of your routing and various export settings then?
ill try to narrow it down. its fairly simple. lets say i have 1 track and only one. i want to export it. i keep the track at 0 db and the master at 0 db. no effects whatsoever on both channel and master.
now, in export dialog box, the first time i choose the export to take place via stereo out (see pic) and the 2nd time i use the channel batch export (pic 2) and choose that one channel. technically - they should be 100% identical but as mentioned before - they are not. this posses a question if i can trust the batch export for lets say 50 channels? i know there is difference between the 2 methods but i cant tell which is “best”. hope i make sense.
did you check that the mixdowns are aligned on the sample level? Even a 1-sample offset will fail the null test. It doesn’t matter that you rendered from the same locator, there may still be a timing difference.
they are aligned. im not doing any moves or actions between both exports
This is not what I mean. I assume that your files start at exactly the same time, the actual audio waveform may be shifted. You have to check. Also, if you set your grid to 1 sample steps, and setup some shortcut keys to nudge events, you can try moving one of the files 1 sample at a time, while playing the null test. It’s quite obvious to hear when files are aligned.
well, it sure worth a try checking. thanks
What is your panning law setting? I’m not sure if the batch export is post pan/fader etc. (have never used it), but standard is -3dB I think, so this may be reflected on the stereo bus.
1 Like