Multi-channel recorded output splitting into stems to send to a DAW

Oh, I forgot that, it worked great on .wav files though, oh well I’ll keep an eye out for a solution. Also, for extra info for all, if using C3 you can switch to .wav or .x64.

Dave

Yes, but .wav only supports up to 4 GB. (Or has that changed?)

Terry

EDIT - Answered my own question:

Hey Terry,

You got my investigative juices flowing on this generic w64 demux/split/convert for Cantabile and I have found a promising home grown batch file conversion solution if you’re interested in the setup, once setup is done things really speed up from my experiments today vs doing each file one at a time in audacity.

Dave

Definitely without a doubt indubitably absolutely seriously yes very interested! :slight_smile:

Terry

Hi Terry,

Ok, good, I’m working on the batch file but need some info from you. The decoder/splitter I’m using needs to know the number of channels in the file before executing the splits and file writes, if it doesn’t have that right it will write 0 size files :astonished: So off hand do you use the same number of channels in your W64 files and if so how many do you use in your “channel layout” ? I can get you going without the fancy interface if you can stand a Command window for the input and then poof the files appear after being split. Given time I can see it getting put in a better wrapper … lol:grin:

Thanks

Dave

Hey Terry,

I’ve built a quick and dirty experimental version that will work on .w64 files - but only with an older version of sox (which does work with w64 files). So get version 14.4.0 of SOX, download my experimental version here and take a look at the original thread here to find out how to configure AudioDemuxer to point to the right version of SOX.

I’ve tried it with a quick experimental w64 audio file from Cantabile - seemed to demux it all right. Can you see if it works for you?

Cheers,

Torsten

1 Like

They are seldom the same number of tracks, as I change up the number of layered instruments frequently, but if I just always grabbed the same number then I could simply always make an 11 gigabyte file like I did today!!! :stuck_out_tongue_winking_eye:

I usually have two mics, four or five hardware synths and five VST’s being captured, and sometimes the old LXP-5 comes into play.

Honestly, the nice thing about sticking with Audacity is that I can make the mono L+R tracks into stereo ones where applicable, and give the tracks names before exporting the stems. I can see whether there actually is anything recorded also (in the case where I checked more items than actually produced sound).

However - I also have the three-channel layout I use every Friday (composite, piano, reverb VST). We could start by experimenting with that one! :slight_smile:

Terry

What’s a tiny bit tricky is remembering which of my checked items were stereo and which were mono, but that is certainly no big huge deal to keep track of.

Only four mono items anyway - two mics, the TX81Z and the ESQm. I think I can keep up with that!!! :wink:

Terry

Hey Terry,

I saw @Torsten posted about his new updated Audio Demuxer, you should try it out first and let me know how it goes. I liked his first one.

Dave

Hi @Torsten,

I took the liberty of trying out the new version of Audio Demuxer after updating Sox and it gave me a group of errors on a 16 channel w64 file I use for testing that was created by Cantabile. Here’s a screen shot that covers as much as possible in one screen. App reports process undertaken, error box shows errors on all 16 channels with the small box as an example of all the others and no files are created or written. Hope this adds to development.

Dave

@dave_dore, you need to use the older version of SOX (14.4.0) with the experimental version. 14-4-2 fails on w64 files, but 14.4.0 works.

Cheers

Torsten

1 Like

Thanks @Torsten I will try it after correcting the version.

Dave

Hi Torsten,

I made the sox adjustment and now it’s working fine here on Win10. I must make some custom profiles now! I have a working DOS line version I made with ffmpeg that is solid but not nearly as elegant. Thanks so much for the update!

Dave

1 Like

Terry, I actually moved over to recording all channels dry and directly in Reaper. But using c3 to send song changes to reaper so that tracks all end up nicely named and all I have to do is a final mix and crop and render.

I can also play back the track during rehearsal instantly from reaper with no other processing. I can also either play it back through the reaper mixer and fx, or send the tracks back through the live mixer so we can hear exactly what it sounded like though foh.

I can describe in more detail if anyone is interested.

I would definitely be interested to hear your alternative method. I use Samplitude, which is surely capable of multitrack grabs. What do you have being sent from C3 to enable track names are already there? That is an interesting feature!

How are you shunting the audio from C3 to Reaper? Virtual Audio Cable?

Terry

By the way, I was glad Cantabile could record within itself as I already am running 6 keyboards and at least 6 plugins (sometimes 9), and OBS is sending out a 720p live stream with two cameras all from the same machine. (Cantabile feeds Voicemeeter via an ASIO ADAT TOSLINK loopback channel in my MOTU 828 mk3 hybrid to generate an OBS-acceptable audio stream in WDM/KS.) So, I was hesitant to tax the system further by adding another piece of software to the burden!

(Sometimes I feel like I am abusive of my poor CPU!)

Terry

1 Like

I found Reaper made almost no difference to the CPU load whether it was loaded and recording or not.

ok, key points as follows

MOTU

  1. All audio channels (around 10) physically come into my Motu Ultralite mk4 which is also the live mixer for the band.
  2. Channels are routed direct into the Motu mixer software (so that if the PC dies, the mixer carries on working)
  3. Channels are also routed (pre fader, pre everything) to audio channels on the PC for simultaneous consumption by C3, Reaper etc. No need for virtual audio cables. Not sure if your Motu is the same.

Cantabile

  1. Binding to send Song program number and bank to Reaper.
  2. Bindings to send midi messages on song load and unload (these are used to start and stop recording). These messages go via a different route that I can enable and disable. This allows me to prevent Reaper auto recording on every song change if I’m not at rehearsal.
  3. Bindings to buttons on my control surface to send to Reaper and enable and disable Record Arm.

Reaper

  1. The Reaper project has all the individual channels mapped and ready to record at any time.
  2. All the songs are set up as regions which are 4 minutes long with 5 minute spacing. Doesn’t matter what order they are in, and it doesn’t have to match Cantabile or the set list on the day.
  3. All regions are numbered - doesn’t matter what number.
  4. Then I use the “SWS Go to Region number” actions and learn the song changes from C3 to the correct region.

In Use
Whenever I change to a different song in C3, Reaper stops recording, moves the cursor to the start of the correct song, and then starts recording again. It doesn’t matter which order the songs are in, either in Reaper or C3. It doesn’t matter if the song overruns or we play it twice. These are just recorded on the timeline, or as 2nd takes.

I did experiment with sending the C3 auto record binding to Reaper to start and stop the recoding, but found that I was sometimes missing the start of songs.

The nice thing is that once the songs are in Regions then Reaper can then render each region as an mp3 stereo mixdown, and wav stems all from one screen.

I use the same recordings to play along to as practice in between rehearsals with my keys track muted. With Record Arm disabled, then reaper simply plays the tracks back on song change. Really easy.

I will say I’ve only just set it up like this - first recording rehearsal will be on Sunday. Before, I recorded all songs in one long session in Reaper. I then used to have to take the 30 to 60 minutes after rehearsal to move all the regions around into their correct places. This way I have a rough recording ready to play back or export straight away, which is often good enough for our purposes.

Playback - MOTU Ultralite MK4 specific
I mentioned that I can play back from Reaper instantly through the SAME mixer channels we just recorded from. I like this because we can go to FOH and check levels and mixer fx from the audience perspective.

To do this means rerouting the specific instrument mixer channels on the MOTU so that the audio comes from the PC audio channel, and not from the physical input jack.

I have 2 batch files assiged to a button that do this rerouting for me using the MOTU API, so that to play back a song we just recorded takes about 5 seconds.

The batch files contain commands like this. Each line changes a mixer channel routing.
curl --data “json={"value":"0:0"}” localhost:1280/0001f2fffe006030/datastore/ext/obank/6/ch/0/src
curl --data “json={"value":"2:2"}” localhost:1280/0001f2fffe006030/datastore/ext/obank/6/ch/1/src

1 Like

How are you routing your VSTi instruments to Reaper?

I only have four hardware synths that are only used for certain songs (though the Wavestation SR seems to be ever-present). I have at least four VSTi’s going, plus the reverb(s) and limiter. I’d have to get those to Reaper somehow, and all my 828 inputs are in use. (Wait, I take that back - I can do a loopback with the ADATs and use those… I already do for some things: ADAT 1/2 feeds Voicemeeter to feed the OBS stream. OK - I have seven stereo outs left, but I often require nine.)

I’ll mess with this a bit also. Probably Samplitude also doesn’t make much of a CPU hit when recording. After all, it was originally written to work in Windows 98!

The 828 doesn’t have the network capabilities of your box, but it does have the ability to be controlled by a remote controller and by TouchOSC. However, I have checked CueMix and it is a real CPU hog, so I would have to simply operate “blind” unless I can hook up my Behringer BCF2000 to it.

But aside from needing to separate out all the stems using Audacity (which isn’t hard and lets me convert to stereo tracks AND name the tracks before exporting stems), Brad’s recorder is working plenty adequately. My only issue was that Audacity only exports in 16 or 32 bit, not in the original 24-bit I’m recording at, so I get inflated file sizes unnecessarily.

Terry

The MOTU ASIO driver provides all the audio channels. C3 sends to a couple of these channels called "From Cantabile I have named the channels from the MOTU’s perspective. There can be 48 ins and 48 outs between the MOTU and the PC…

Within the MOTU routing I send those channels back straight out to Reaper via another couple of channels called “Mix Keys”. This routing does not go via the Motu mixer, it is simply in and out. Not shown, but further down this matrix screen I also send this same audio coming from C3 into the MOTU mixer as 2 channels.

Then I connect Reaper to those 2 channels, as well as to the other MOTU channels that are routed from the ADAT input. I actually ALSO connect C3 to all these channels simultaneously with Reaper, as at one point I was recording with C3 and Reaper at the same time from the same channels.

Does this make sense?

1 Like

The 828 lacks this routing matrix, though it can be “tricked” by using a TOSLINK fiber cable sending the ADAT outs to the ins producing up to 16 loopback channels. I do that now with a single fiber optic loopback for eight channels (of which I’ve only been using two) but I have another fiber cable to employ the second ADAT port set.

This certainly makes sense and seems feasible - removing the need to extract stems at all! I’ll try it this week, for sure.

Terry