How to render track-level stems exactly as they sound in the mix?

[med3.21]

Hi there

I have a finished song and have been asked to provide stems. The track contains a few groups and subgroups. There are things going on like the Signal Follower effect creating a “ducking” kind of sidechain on one of these subgroups whenever a kick drum sounds. And there are master- and group-level compressors/tape/etc effects that are responding to the entire summed signal going into them.

I want to end up with a set of stems representing each track (not group) which, when placed into a DAW, will recreate the exact sound of the original Renoise tune. I am not interested in having the groups and subgroups rendered as their own WAVS, as the end user is only interested in the individual tracks within them.

If I select Render to Disk and tick “Save each track into a separate file” to extract the stems, it gives me separate WAVs for each track, each group and each subgroup. But the individual track WAVs are dry whenever those are part of a subgroup. For example, when I look at the WAV for the rendered track representing the bass guitar, the ducking effect is missing. So when I assemble the track-level WAVs into a DAW, it recreates the tune without any of the group-level FX or other track-dependent things like Signal Follower on it.

Then I found this ace tool by @dltfm :

The Deliverer

I thought this would do what I wanted, but it seems to work by soloing each track as it goes and rendering a WAV of each. This ignores the Signal Follower when generating WAVs, as the Signal Follower must appear on a different track to the one you’re soloing, so will be inactive and not producing its required control signal to do the ducking. Not only that, but master FX like compressors/tape FX will end up sounding different as those effects respond in a different way when presented with the summed input signal of everything on the tune versus the one track being soloed at the time. So when I end up with a set of WAVs extracted with The Deliverer, it’s closer to what I need, ie it has some of the effects “burnt in” to each wav, but seems to omit ones that rely on other channels that would be muted at the time, and tape/compressor effects do not sound the same. Therefore the set of WAVs I end up with still do not sound like the original tune when summed together and because master effects like limiters/compressors act differently on solo signals, the assembled WAVs will exceed the peak signal.

Now I’m experimenting with setting up a send for each finished track I want and seeing if that will conjure up a set of faithful stems, but that seems to be its own world of pain too.

Is there a way/setting/tool that will generate track-level “wet” WAVs that, when summed, will sound exactly like the finished tune?

[eretsua]

If you applied the effects to the (sub)group then it is normal that the individual tracks within the group are unaffected by those effects.

AFAIK there is no easy solution for this. You’re either going to have to copy the effects to each individual track. Or try to do some workaround with solo’ing and lowering the output volume of the kick to -inf. So the Signal Follower still happens but the kick isn’t heard. However, I don’t expect you’ll get it to sound exactly like it does now.

[dspasic (senseiprod)]

@med3.21

If I select Render to Disk and tick “Save each track into a separate file” to extract the stems, it gives me separate WAVs for each track, each group and each subgroup. But the individual track WAVs are dry whenever those are part of a subgroup.

you are asking for something impossible…

if you concat 2 audio streams it’s a totally new audio stream, and apply effect which depends on merged audio source (newly merged audio stream)- which has no relation to separate audio sourced - and therefore you cannot have two separate wet signals in this case…

This is not related to specific Renoise functionality, since in any DAW this is what it is… It’s about your understanding what goes on

you have summed master (or group) signal which is processed as a single source… This means that the signal is no longer multi-track (as separate tracks), and therefore you cannot create separate tracks/stems out of single audio source

[slujr (•MrZensphere•)]

I would just put a gainer set to -inf after the signal follower(s) and then render those soloed “sidechained” tracks out individually with the now silent sidechain impulse track soloed as well. capisce?

it’ll be a little tedious as a workaround, but should yield the result you’re after

[med3.21]

Thanks, all. I figured out the perfect solution! Here goes:

The key to the puzzle is MUTING (left-click) the tracks you don’t need, rather than soloing (right-click) the one you do. That way any other tracks that influence the one for which you are creating a stem (eg. a kick drum that controls the Signal Follower effect on a relevant subgroup) are still working their magic, albeit silently. When you use the solo function, this behaviour does not occur.

I just created a perfect set of stems that, when summed in a DAW, sound EXACTLY the same as the master output render. For example the bass guitar stem on its own has been affected by the sidechain kick exactly as it would have been during the full mix, and the ducking effect is audible when I play it back on its own after rendering. Creating a stem by soloing the bass track would not have achieved this.

It’s a little fiddly where groups and subgroups are involved, so I thought I’d include some examples below (the required stem is the track marked in red each time):

In this one, I needed a stem of just the bass guitar. I had to mute the Drums group, the ungrouped track called Clappy and the Non-[basses] subgroup on the right, as well as Synth[Bass] ie the other track in the Basses subgroup. Also sends S01 and S02. Essentially mute the highest level of group that doesn’t include the stem you want, mute the non- required ungrouped channels, and mute any other tracks in the group containing the stem you require, as well as unused sends for that stem. (Phew!)

In this example, I needed to create a stem of Hats, the hihat pattern. I muted all the non-Drums groups at their highest level, as well as Clappy (the only ungrouped track), and the send returns S01 and S02. Within the Drums group I muted each non-Hats track.

For this one, I needed a stem of Strings, which has a send return on track S02. It’s the same principle as before, ie mute the Drums group, mute Clappy, mute the Basses subgroup, mute everything else in the Non-[Basses] subgroup (Hypnot…, Piano, Chord, Wurly, etc) and mute S01.

At the end of this somewhat laborious but entirely logical (and potentially programmable - @dltfm?) process, I had the most perfect set of stems I have ever created on a DAW. The eureka moment was grasping the difference between muting unwanted tracks and soloing the required one. Hopefully this helps others!

(Edit: I should mention I am on Windows, in case the above doesn’t work on Mac/Linux, though I don’t see why it wouldn’t.)