Do you mix stems of virtual instruments?

This is a workflow question. I’ve come so far as to separate tracking and mixing (more or less). But I do consider the art of making an instrument sound good is to some extent part of the tracking process. By the time I get to the mixing stage, I’ve got quite a few plugins running, taking up quite a bit of my computer performance. If I render the mixbus, the program sometimes crashes, or you might just get a bad recording of some of the instruments. What I often do is render individual tracks to stems and mix those. But then I find out during mixing I don’t like some aspect of the stem, and go back to unfreezing the stem. It can keep you busy for hours…
Anyway I was just wondering if you guys and gals mix audio files or keep your VST instruments, amps and such in the midi domain till mixdown? And if not, at what stage do you render your files?
I’d like to become more efficient at this :sweat:

I don’t render or freeze. If the load gets heavy (which is rare) I outsource it to another computer. The one exception is that I do render autotune. “Bounce in place” or “render in place” as its called on Logic, Cubase, and ProTools 12. But only because autotunes can be unstable, inconsistent, and they cause a lot of latency.

I also don’t mix anything down until everything is done at once. I leave everything as-is in the individual tracks. My machine is an i5 with 16 gigs and according to Mixcraft, which has a little resource meter down in the corner visible at all times, I’ve never come anywhere near using all my resources. The highest numbers I’ve ever seen are on the order of 25% and then only in short bursts. 15% is very typical.

MC’s resource meter includes both the percent used by itself, and the percent in use in the entire system. The system resources are above the Mixcraft resources by only 5-10% typically, because I usually have nothing else running (a web browser, once in a while). Evert, do you have any other processes running in the background on your rig? That might be contributing to the issues you’re experiencing. I build my own computers so I know exactly what’s on there, and there is none of the bloatware that commercial boxes typically come larded with. Might want to look into that if you haven’t already…

I also have a very powerful machine now, so don’t really need to freeze anything. I often freeze the drums though, while I work out everything else. Feel sorry for the drummer, I suppose… using Kontakt, I also purge the samples before freezing…

I commit to all my sounds before I start mixing them. That means rendering/freezing/bouncing any virtual instruments. Sometimes I will keep amp sims in line, but they are the next thing to be rendered if the CPU starts wheezing.

1 Like

I’m not quite sure. I bought the computer from someone who built it himself (i-7 chip from 2014). I don’t think there’s much running in the background, I turned quite a lot of stuff off. It could maybe do with a bit more RAM (8 gigs at the moment). I run it al in Reaper, which also lets you know how much resource is being used, usually no problem, but in practice I do run into problems way before I reach any limits of the computer. I don’t know why. One strange thing is that I can’t run Reaper in normal mode: it has a huge latency. I run it in a sort of USB plugin mode designed to take your setup with you on a USB stick. It works, but maybe only to a limit. I’m hoping a friend with more computer skills will help me reset the whole thing so I can run Reaper in a normal way, and maybe not run into limitations with RAM. Another thing I want to try is run a lot of stuff from a SSD disc. But I don’t dare do these things by myself, afraid I’ll make a mess of things…

But my question was also a more principal one: do you commit at any time to something which you think: “this is good enough for now, lets get on with it” and you render it to a WAV file. I would have guessed that the pro’s would do that. But judging from Jonathan’s reply (who is certainly a hell of lot more pro than I am) he keeps his options open till the last moment. Well why not, if your computer lets you. One exception is Autotune apparently. Well I don’t use Autotune, though I do use the similar Reatune now and then on the vocals. Might be worth looking into if that requires rendering!

Understood-- and this is what I do as well. There is zero reason for me not to. I like having all options open at all times, and there is no cost in performance or anything else.

Weird that Reaper chokes on what you’re throwing at it… doesn’t sound like it should. Good luck!

So it sounds like you’re using the Reaper “portable” install. That is handy for moving between machines and setups, but I would think the full install on the computer has some advantages. I can’t imagine that the portable install would use less resources on your computer than the installed version … assuming comparing the same project with VI’s and other plugins. The CPU load and RAM requirements should be the same or similar I would think. Are you setting your latency buffer in Preferences (Audio | Device | Request block size)? Maybe your full installed version is corrupt or something? You could try uninstalling and starting from scratch, just back up your Reaper config files (hopefully those aren’t also corrupt) before uninstall.

I learned from Eddie Kramer that he has done this for years and years, way back to Jimi Hendrix and Led Zeppelin. And he’s still at it! (the last I knew anyway) It sounds like Coldroom (Andrew) is saying the same thing too. If you have too many options or ‘exits’, it’s easy to get overwhelmed or distracted. Committing to sounds and forging forward can be scary and unnerving at first, but with practice you are actually building your ability to know which sounds you like and will work in the format of the song. It’s a path, and maybe works best for people who consider themselves “intuitive” in their approach to mixing (that feeling in your gut that it’s either working or not working). Someone who takes more of a “sensing” approach may find the opposite works for them. These are personality characteristics, so if you observe yourself you may come to understand which is most helpful to your process.


Interesting. I render midi instruments once the “performance” is what I want.
After all, If I’d got a pianist to do a session, I’d have to mix with his performance and it wd be expensive top keep calling him/her back for a new session.
But actually, it’s because I’m a tweaker. If I can change something I will and trh temptation to tweak the midi gets in the way of the ongoing mixing.

The main reason I render out my samples is because a lot of times when you start playback in the middle of a phrase, it doesn’t play back right, due to keyswitches and automation of CC values and whatnot. Plus it makes it faster to load a project if there are lots of big sample libraries on it.


It depends.

I use a fair number of outboard synths – some older like a JV1080 and a TX81, and some newer like a mopho. The midi is in my daw of course, goes out, and then the audio comes back in to another track. With those, once I am satisfied with the sound in general compared to other things, I tend to “print” that audio then disable the originating midi track.

My projects tend to be comparatively simple and I don’t render to save processing power so much as to remove the distraction of the outboard synth as I move forward.

Non midi stuff such as recorded vocals or guitars can be both easier or harder.

First off, I do the unthinkable in that I “print” certain effects while recording – most particularly expansion (to reduce background noise as well as hum etc) as well as mild compression. I just use a cheap dbx 266. This means I most likely won’t need to compress a track.

By easier, I mean I usually will have only a couple of plugins anyway, because I tend to record live stuff through an expander/compressor in the first place, most often a track will contain little more than (maybe) a noise reducer (e.g. reafir) and a reverb though in some cases I find an eq useful especially for acoustic guitar. In these cases I tend not to freeze or render the tracks because they aren’t either distracting or intensive.

By harder I mean that sometimes I set up something elaborate. Like for a mono acoustic guitar I might pan it 45 degrees right, and then have the reverb for it panned 60 left on a separate channel. Where that can get complicated is eq both before and after reverb, and I’ll set up a partial autopan fairly wide on the reverb channel and fairly narrowly on the original track. (By partial autopan I mean that goes to yet another track which just has autopan and is separately mixed).

You get the idea. For something like that, it not only can be processor intensive, but it can tempt me to endlessly tweak. Instead I just fiddle until it sounds good, render it to a track and then mute and hide the originals. I can always come back and mess with it later if there is something wrong. For something like acoustic guitar you’ll never hear that it is autopanning the effect.

For midi soft synths (e.g. sylenth) and drums(ez drummer etc) I seldom bother to render them. I haven’t really seen any processor intensity issues.

I think what you are reporting indicates something is awry – it could be your driver is set for too low a latency. (remember you can always line things up perfectly later) or there could be a bad driver gobbling cycles. There are also some badly behaved plugins out there. Likewise the “portable” install can be quite slow as Stan mentioned. We have similarly powered systems so you shouldn’t be seeing performance issues.

So something ain’t right.

Well that is true… I really tried everything short of a complete re-install of all my software (including windows). I was helped by people on the Reaper Forum, but nothing could be found. Until I stumbled on Reaper Portable install (yeah that’s what it’s called) which for reasons that I don’t understand works fine. You can record at quite low latency without pops and crackles and not run into any problems till the track count (with all sorts of plugins) reaches around 40-50 tracks. Particularly my orchestra plugs. By that time I need to render some tracks (which you can always unfreeze, but if you go back and forth it takes quite a bit of time). I’m thinking of running all of my software and samples from an SDD card (it’s been lying waiting to install for months already), so I have to reinstall windows anyway. Only problem is that I don’t have a Windows install DVD. I assume you need one? Anyway this is the sort of stuff I need someone who’s more computer savvy than me to help me. Preferably in a way that I don’t need to reinstall all my plugins (I know, I have way too many…). Does anybody know what the install of Reaper Portable does differently from normal Reaper? Somehow that’s related to latency issues.
Any ideas would be appreciated.

This talk of latency made a neuron fire. Forgive me if this seems insultingly obvious*, but do you relax your settings to a long latency when not actually tracking? That is, get it down to that 3 to 4 ms while actually tracking parts, but then boost it back up to 50 or so for all the mixing work (which for many of us I’m sure, takes many times longer to do than the actual tracking does). When you’re not asking your rig to be low-latency, it can handle WAY more.

Sorry if you’ve already done this but I thought it worth mentioning…

*When I was in grad school I earned the title Master of the Obvious. :grin:

Only if I start noticing issues. Sometimes I can make it through a mix and just keep it on low latency, other times super low latency causes some issues. I don’t make a habit out of it making sure I’m at a higher latency when mixing. Once I get above 1024 sample buffer, I don’t really notice much benefit.

Same here, but I get a huge benefit by shifting from ~150 up to that level… just part of my SOP now to leave it at a high number any time I am not actually recording something, and put it right back as soon as I finish.

The difference between a “regular” reaper install and a portable one only lies in two areas.

The first is that a regular reaper install puts certain data in the users home/application data folder for reaper and the program itself elsewhere whereas a portable install puts all of it in the same folder. Reaper doesn’t store stuff in the registry, etc making a mess. So the only on-disk difference is where the data is stored in terms of folders.

The only downside to a portable install in this respect is that a regular one allows different users on the computer to have different things going on in reaper, whereas the portable install is really a one-user thing.

The second difference is storage medium. Although you CAN do a so-called portable install in a directory on your internal HDD, most often portable installs are on thumb drives, SD cards etc.

Okay … if portable works okay but regular doesn’t … lets break down a couple of things.

HDD – it’s transparent to us because we just put stuff on and take it off and it works. But there are three distinct times associated with it: seek time (time to find a particular piece of data), read time, write time. If a piece of data isn’t contiguous, then for each piece of data a seek has to be performed before reading.

Pretend you have 20 tracks. For simplicity we’ll make them all midi. Each of these has a file associated with it in reaper. Now, reaper will try to hold all of these in memory so it doesn’t have to do things on the drive, but depending on how much memory you have, how much is used by other stuff etc. But a standard HDD has around a 15ms seek time, and will handle about 100 I/O operations (reading or writing) per second.

That probably seems awfully slow when running a processor that executed 3 BILLION operations per second … but that is the nature of a disk that spins around in circles at 5k-7k rpm and the head can’t read or write something until it gets to the right place.

So a standard mechanical hard drive can put some serious bottlenecks when you start running a lot of tracks.

An SSD by way of comparison has a seek time of 0.1ms – 150x faster – and in terms of I/O per second for the same size smaller file is about 53x faster.

Which brings me to things on USB. USB3 or similar memory-based drives can in practice be dramatically faster than an internal rotating hard disk. USB2 … not so much. That can be about the same. And 1.1 … is actually slower. But if you are running USB3 or a similar technology for a memory based device for a portable install, it will substantially out-perform an internal spinning drive.

Is your internal drive a spinner? If it is you might consider migrating to an SSD?

Also gonna mention latency settings make a big difference on the processor side of things. But since there is a big difference between portable and normal install here, I am going to guess the problem is a hard disk bottle neck.

Same here too. I usually go down to about 96 (or is it 69) to track, and (if I don’t forget) go back to 1024 to mix.

1 Like

Wow, thanks for the clear explanation Brutus. I get what you’re saying and its pretty consistent with my intuition (not that that is any sort of useful reference :sunglasses:). But it is the reason I bought an SSD card. My portable install is running on the hard disk (on my virtual C drive. I have samples, reaper files and WAV files on the virtual D drive).
Your analysis also triggered the thought that maybe my computer is not getting full use of the 8 K of memory for some reason? Can I check that somehow?

Windows has a performance monitor in it – I think it is just one of the tabs in the task manager (hit ctrl-alt-delete, select task manager, then hit the performance tab).

If you poke around a bit, you’ll notice that you can see how much of your ram is in use or free, and which applications are using ram, and how much they are using. I can definitely say upgrading my ram from 8Gb to 16Gb made a big difference.

Reaper itself is pretty gentle on ram, but plugin samplers especially tend to load an entire sample library at once into ram so it can run in real-time.

So I think you may be on to something here! Load up your reaper and run one of your mixes and see how much ram is being used – it may be that a ram upgrade will make it run better!

Disclaimer: I’m a computer geek, but my specialty is Linux rather than windows.