Exporting/mixdowns

[karmacazee]

Ok, so I'm currently working on a new project and I'd like to get the mixdown spot on so that it can be mastered.

One thing I normally do is bounce down all my tracks to audio idividually and mix them as WAVs (to save CPU and to have copies of everything just in case the computer goes tits up).

My question is... when I'm exporting my individual tracks out of Ableton how should I do it?

I mean, should I try and get the maximum level out of every track, or should I bounce them as they are, as the mix I have right now is pretty sweet. Also, Which export options should I choose, I remember reading somewhere that Ableton has all these dithering options on the export screen, but I have no idea what the difference between them all is...

Or should I just skip this process and get a good mixdown with my MIDI tracks and my CPU meter running almost red...


Edit: Also, at the moment, it's really hard to get the levels right in my project as it is now because of all the automation I have going on...

[FSTZ1]

Or should I just skip this process and get a good mixdown with my MIDI tracks and my CPU meter running almost red...

yes

edit: you can also upgrade your ram for almost nothing nowadays

[Sharmaji]

i always bounce out audio for every track, and generally run each track through some hardware chain (pre/compressor/eq) to really get some hair on it. a couple of reasons:

to my ears, it sounds better.

i don't want to think about mix issues when i'm in writing mode. the only mix thing i'm concerned about is that nothing's in the red. i'll slap on a compressor or eq in that state to give something character, but i'm not worried about how things fit together sonically in the writing stage-- i'm concerned about everything sounding bangin' on it's own terms, and the arrangement.

i like to use some pretty hefty plugins when i mix, and doing that with instances of massive, sculpture, albino, battery etc-- just gets to be the coughing, sputtering mess.. and it doedsn't have to be.

lately my mixes have been going through a more organic move from midi to audio-- once i have something set, i'll commit to it, do it's hardware bounce out if that's what it needs, bring the audio into a new channel and disable the midi channel that was running it. in a way it lets me work faster and commit to an idea.

for plenty of folks, it's a linear process, but for me i know that i write more courageously (and faster) if i'm not worrying about compressor ratios, etc, and i stay focused on sonics if i'm less concerned about arrangement issues. splitting it up lets me make better decisions.

[Jah Billah]

Bounce everyting to audio at 96/24 & mix it all up.

[karmacazee]

I think a comprimise is in order here... the tracks that are heavily automated wil just be frozen, and the ones I'm happy with shall be bounced.

Do you bounce them as they are, or do you get a good level from each solo'd track?

And if someone can fill me in on the dithering thing...

I'll (obviously) post my mixdown when it's done

[Sharmaji]

I think a comprimise is in order here... the tracks that are heavily automated wil just be frozen, and the ones I'm happy with shall be bounced.

Do you bounce them as they are, or do you get a good level from each solo'd track?

And if someone can fill me in on the dithering thing...

I'll (obviously) post my mixdown when it's done

work at 24 bit and bounce at 24 bit and you won't have to worry about dithering. dither is essentially low-level noise that hides the junk introduced when audio is truncated from a higher bit length to a lower.

i bounce things as they are, save for 'verb, unless the 'verb is the exact, perfect, most amazing verb for the sound ever.

[karmacazee]

work at 24 bit and bounce at 24 bit and you won't have to worry about dithering. dither is essentially low-level noise that hides the junk introduced when audio is truncated from a higher bit length to a lower.

i bounce things as they are, save for 'verb, unless the 'verb is the exact, perfect, most amazing verb for the sound ever.

Thanks!

at the speedy reply too!

[macc]

^ 32-bit float is essentially 24-bit resolution with 8 bits for +0dB information, so unless everything in your mix is clipping then 24 is perfectly adequate (ie wonderful).

My 2 penneth

If you're going to be doing as described and bouncing out stems for the final mixdown then exporting at 24-bit at unity gain (ie, the level they are in the mix) is the way to go. This means that if you have a shaker or something that is super quiet in the mix, peaking at -20 or so, it is still captured at better than 16-bit resolution.

Turning up the levels of the tracks to try to maximise the level of the individual parts is, frankly, a waste of time. You have to either take loads of notes to turn the tracks back down to the right level again to rebuild your mix, you run the risk of clipping (!), and you aren't really gaining/retaining any quality in terms of resolution - especially as you'll be turning them down again anyway. Better to keep the master fader where it is, reimport all the tracks, then reset the master fader to zero if it wasn't (tut tut ) and away you go.

Of course, the caveat here is if you're doing the analogue thing for some 'hair' as TeReKeTe put it. That man certainly loves his hair

[Brisance]

CPU meter on red during mixdown? Jesus, do you run tons of vsts or is ableton just a cpu hog? With FL, my cpu doesn't even have a role and I take ~300MB of RAM maximum.

[rob sparx]

Bounce everyting to audio at 96/24 & mix it all up.

I've bounced down premasters at 96/32 (which always seems to leave engineers scratching their head looking baffled for a bit trying to convert it at the mastering stage) but is there actually any point in bouncing down every single channel in the tune at 96? I mean some of my tunes have 60+ channels so your talking about using a shitload of HD space - the only way your going to benefit from the extra resolution is if you do plenty of processing after the conversion and and chances are that you might find you need to add extra layers to make the mix work (I'm always adding extra channels even near the end of mixing down a tune) that means converting any new samples to 96khz just so you can hear them at the right pitch as (at least on my soundcard and DAW) you can't mix 44.1 and 96 together without the 44.1s being pitched differently.

I can imagine if your doing live music or music with lots of live channels 96khz is a good resolution to work at but for music made with 44.1khz samples and loads of channels its a bit overkill isn't it?

[paradigm_x]

i always thought that, but apparently vst synths/samplers etc all work better at higher frequencies, due to reduced/improved aliasing.

I read elsewhere also to work at 24/44 to keep cpu usage down, but bounce out at 24/88 which is what ive been doing.

However, ive read since, from a reliable source, that this is worse, as cubase's SRC is shite, and youre better bouncing at what you wrote it at, SRC with a good one eg voxegno rbrain, then post processing, then SRC/resampling down to 16/44.

Not tried the differences TBH but i was getting a bit fed up of 88k files anyway. My soundcard glitches everytime it gets a new samplerate, and sometimes gets stuck (nned a new one asap).

Think ill just stick to 24/44. the '24' part is far more important than the 44 bit anyway.

[whineo]

My question is... when I'm exporting my individual tracks out of Ableton how should I do it?

I mean, should I try and get the maximum level out of every track, or should I bounce them as they are, as the mix I have right now is pretty sweet. ...

Freeze > Flatten

[macc]

I read elsewhere also to work at 24/44 to keep cpu usage down, but bounce out at 24/88 which is what ive been doing.

However, ive read since, from a reliable source, that this is worse, as cubase's SRC is shite, and youre better bouncing at what you wrote it at

100% Avoid unnecessary SRC at all costs IMO (well... see below). Working at high SRs is one thing, conversion is soooort of different, to my mind anyway.

SRC with a good one eg voxegno rbrain, then post processing, then SRC/resampling down to 16/44.

If you must, then this. My jury is still out on the improvements of upsampling for ITB proessing etc prior to going out to analogue, then downsampling to 44.1... I appreciate the benefits but 2 lots of SRC makes me go all a bit queasy, even with r8brain/SOX. But then lots of plugins oversample for processing anyway, so I still can't decide. I've had zero comments back from customers whichever way I have done things tbh...

One note, if downsampling after limiting besure to leave a little (0.2dB) headroom as you'll get small new peaks when downsampling (which you can then limit at 44.1 if you so wish) as a consequence of the downsampling.

Think ill just stick to 24/44. the '24' part is far more important than the 44 bit anyway.

[karmacazee]

CPU meter on red during mixdown? Jesus, do you run tons of vsts or is ableton just a cpu hog? With FL, my cpu doesn't even have a role and I take ~300MB of RAM maximum.

Well, not quite in the red, because that means crashy crashy, but my latest project is running quite hot. Kontakt 3 seems to be a bit of a CPU hog on my laptop, especially if I load more than one instance of it.

Plus I've automated many parameters on nearly every single track, so this one's a monster.

Normally it never goes beyond around 25%, but I think my projects have gotten a little more complex.

[grow]

I like to make multiple copies of my tracks with different names as I start to flatten. I sometimes work with other producers who are anti bouncing down as they say the more renders the more digital artifacts...ever since 7 I honestly haven't been able to tell the difference so I freeze and flatten. I also feel you have more control over the audio for making cool edits and variations during the final stages of production to make the track more interesting. I built a beast I7 so I never have problems with cpu. If you are having cpu issues and more ram just isn't cutting it and you are a stickler for analogish sound I recommend freezing during production and then unfreezing before the render to minimize the number of times the sounds are processed through the rendering engine.

Live 8 should make no difference as the rendering engine is supposed to be non-degrading. Anyone got an observation on this?

[Sharmaji]

Of course, the caveat here is if you're doing the analogue thing for some 'hair' as TeReKeTe put it. That man certainly loves his hair

in truth it's the only reason i'm involved in music at all. that and the huge sums (HUGE!) i make writing, mixing, and djing dubstep

[karmacazee]

Cheers everyone, mixdown is going well now (I've never paid this much attention to a mix before, lol. There is a lot going on in this one, so much for the minimal approach...).

I decided not to bounce in the end. Instead I froze a lot of things and routed almost everything through several return busses, and mixed them from the busses. Much easier than tweaking all the automation lanes (there's a few fades and that), and couldn't be bothered to wait for everything to export

One little question I have now though is regarding Stereo/mono.

A little example - My bass has been mono'ed using the built in monoiser in Ableton. I then duplicated the track and added some stereo width to it so that it really stands out on stereo speakers, but won't be lost because the mono track is still right there in the middle.

I've now routed them both to the same buss, but will this defeat the object? Will I lose the mono? Should I route them to separate busses and whack the monoiser on it again?