Trying the Zynthian OS before buying a kit, please help

It seems to me (and I think it’s a great idea) that some, maybe even most, of the functionality you’re looking for with the “ReFace” units and the vortex, could be provided by a “driver” - not a driver in the normal sense of the word, but a Zynthian driver:

Here’s a good example of what you can do:

Keep in mind that this is a new feature for Zynthian, so it is not available in the current ‘stable’ release. It will be in ‘Oram’, which is available, but in a non-‘stage ready’ version - meaning it’s still being worked on.

1 Like

This is a long thread and I skipped a bunch, but does a 4-operator Reface even sound as good as Dexed?

Seems like a lot of effort is (in the case of this particular device) being put out for an external device that is less sonically rich than the comparable zynth engine.

I cannot say that I’ve compared Dexed on the zynthian to the Reface, but I’ve compared Dexed on other computers to my real DX7s and I really couldn’t hear much difference.

edit: possibly the Reface DX got an engine upgrade since I last looked into it, it’s been a minute since I looked at that lineup.

I might be inclined to use Dexed for some things over the Reface DX, but using Dexed would mean I couldn’t use the Zynthian’s audio in the same song for something the Reface DX can’t do, like Salamander Piano for instance. So having it gives me more options.

It 's a surprising (and totally wrong) conclusion that I don’t know where do you get it. Zynthian is fully multi timbral, @BenMcLean .


Oh oops, I actually do know that. What I said there came out wrong.

I’m actually planning to do multi-track recording with my Tascam 16 mixer whenever we have a live show. With the Reface DX (that I already owned before I’d ever heard of Zynthian or Dexed) I can get the FM synth sound on a separate channel from the Zynthian’s channel on my Tascam 16 mixer.

So I might used Dexed for some things if I’m not using the Zynthian for anything else on that same song, but if I have a need for another sound, like Salamander Piano, then I’d be inclined to use the Zynthian for that and the Reface DX for the FM synth sounds so I can get both those on separate Tascam 16 mixer channels.

That’s what I meant to say. I do know that the Zynthian should be able to produce multiple sounds at once. (at least in theory) Sorry, I just kind of said that wrong.

That makes sense - wanting to record different sounds on different channels. Be aware that Zynthian does support multichannel sound cards as well as multichannel audio and MIDI recording. These are features are are not often used but can turn Zynthian into a multichannel sound desk with multitrack recording. I use mine this way but am yet to load test it to benchmark how well it performs. Maybe something I could do tonight…

1 Like

I had thought about trying to connect the Tascam 16 mixer via USB to be the audio interface for the Zynthian instead of the Zynthian’s built-in audio interface, but I seem to recall reading that Zynthian OS was prone to crashing if external USB audio gear ever gets disconnected, with a lengthy setup process to restore functionality after reconnecting it. That seems like it would be a dealbreaker for that plan.

BTW, it is a common misconception that the Reface DX is emulating the DX7, where in reality it’s actually closer to the DX100. While it has less operators than the DX7, it also has more choices for connections between the operators than the DX100, so it isn’t an exact replica of anything.

Update on my attempt to record 8 stereo audio feeds.

It worked! I recorded for 43 minutes at which point I noticed the file was 3.5GB which is approaching the 4GB file size limit of FAT32. So I stopped the recording and loaded it into audio player. It played and I could select each of the 16 tracks individually as well as listen to the stereo mixdown. I observed these issues:

  • Xruns: There were a few xruns but all were short (reporting 0.0us) without noticable impact on the audio. They were concentrated on a particular time which I think corresponds with me dicking around with something.
  • Audio Player could not display the waveform and incorrectly shows “File not loaded”. This is due to an overly simplistic waveform display mechanism. (We can’t fit 4GB samples into 3GB RAM!) I will work on a better waveform display.
  • Slowing down playback too much caused an unrecoverable exception. This doesn’t surprise me. This is very new functionality and I have a fair idea of the cause - again an overly simplistic approach to time stretching. (I think the buffers automatically resize and this is not handled gracefully.
  • The multichannel file was imported into Ardour okay but I didn’t change the import rule so it put all 16 tracks in a single Ardour track. I remove the track and reimported with the correct setting to get 16 individual tracks. (I wish it offered to import as stereo pairs.)
  • There is 24ms between the synthesised drums and the recording of the same track sent out to direct output, looped physically back into an analogue input and recorded through a chain. (44100 sps, 2x256 frame buffers through a Behringer 1820 connected via USB.) That is just over 2 processing periods which is cool - one period to get the audio out and another to capture it back in. Of course I could use latency correction in the DAW to correct this but it was a contrived experiment.
  • Audacity also imported the multchannel file. FYI: Recordings are 16-bit signed PCM WAV files at the native sample rate of the Zynthian which is pretty portable but maybe we could offer an option to record in any of the other supported formats (WAV, AIFF, FLAC, OGG, MP3, AAC, etc.). Audacity stuggled more than Ardour on my old laptop. (I gave up on the Chromebook - too little memory.)
  • I stopped before hitting a file size limit but had forgotten that we use jack_capture which automatically splits files to avoid this issue.
  • It would be advantagous to be able to record mono tracks. If there is a mono source it just wastes disk space and processing to record stereo.

Overall I am quite pleased with this experiment. It shows that Zynthian can be used as a multitrack recorder, mixer, sequencer and multi-instrument softsynth simultaneously to a reasobable quantity of channels.

Zynthasia uses these chains:

  • LinuxSampler->Roomy
  • LinuxSampler
  • Helm
  • Arpegiatore->Helm
  • LinuxSampler->Roomy
  • MDA DX10->MDA Delay

That’s true. Hopefully with the move to bookworm we can look to leverage pipewire in the future instead of jack which will solve this. I have a fiendishly complex workflow to connect my yamaha VKB-100 I’d like to solve.

It is true that Zynthian is predicated on jackd running and that jackd depends on a soundcard to run. If jackd loses its soundcard then it has a hissy fit and doesn’t seem to recover. Worse - if jackd goes down then oit can prove challenging to get other applications to reliably reconnect, even if restarted. This may be just a lack of investigation from our side - after all, Zynthian official kits have an integral soundcard physically connected to the RPi.

But that is not to say that USB audio is all bad. As long as you protect the physical connection between the Zynthian and the soundcard it should stay intact. Indeed my multichannel setup consists of an enclosure that houses everything to avoid external USB connection. (That is provided by a dedicated powered USB hub mounted on the enclosure edge.)

Many people use USB audio for live performance including laptop users who often have little alternative.

I remain sceptical of pipewire. I would love to be convinced that they have accessed the magical pixies and implemented functionality without adding significant overhead… but I am a scientist, not a fantasist!

Why aren’t we using exFAT? Linux kernel supports it now.

I used whatever the USB drive was formatted as. I didn’t check before starting the recording so thought I should interrupt the recording to get some results without the large file issue interferring. I will do another test with a longer recording. Also, large files are unwieldly. It takes a long time to move them, open them, etc. It will be a good test to see if a DAW will import files and position them consecutively without discontinutiy.

It has always been the stable handling of jack that really impressed me in this project. I have always been a “Cubs fan” level advocate of Linux for musicians and the couple of times I’ve used it onstage I always found someone to show it off to, but when I am in a more grounded state I have always been aware that jack is a serious PITA and not stable.

Pipewire seems to just imitate it perfectly and be also quite stable. The only issue I have identified with Pipewire was when I was messing around with snapcast recently - a client running on pipewire was out of sync with nearby clients running raw alsa.

edit: for those who have not grown up within a day or so’s drive of Chicago, Chicago has two baseball teams, a Sox of some colour or other, and the Cubs. The Cubs are… well I’m not a baseball fan so I can’t really offer real insight, but, they have not won a World Series in over a century, and they have legions of maniacal fans who will never, ever, abandon the team. If you visit Chicago you will see many people walking around in more than piece of Cubs clothing throughout the day. I do not recall seeing many Sox hats or Jerseys. I’m sure they were there. I am not a baseball fan, but it’s hard not to love the Cubs.

I just can’t seem to get past this issue here being rejected. Without some way to automatically test for missing hardware, I could end up loading a snapshot with the presets for a song to be sent out to all my keyboards and think they’re ready to play only they’re actually not. That’s a dealbreaker for relying on the Zynthian for presets and MIDI routing for live performance because there really isn’t any way to work around an inability to tell if it’s working. When playing live, you can’t pause to manually test before each song: the test needs to be automatic.

Hi Ben.

I have been using two zynthians for live performances for two years now. I don’t use the MIDI routing as you do, but I use several MIDI controllers which are essential for us.

I make sure everything is working before starting the show, I wouldn’t have time to even unplug the controllers during the performance as we can call Z53 subsnapshots several times per song. And it works very reliablly. I wouldn’t use it otherwise as we are a professional orchestra.

I don’t mind the exact controller I plug in, I could even use different ones. For instance I can use any keyboard as long as it is configured with the right MIDI channel.

This kind of check list probably could be implemented but it wouldn’t solve your problem for live performances, IMHO this has to be checked previously,



I did not see that post before.
Thank you very much for the mostly positive report about multitrack record/playback, it’s really interesting.

1 Like

Dear @BenMcLean !

Yours is a very special use-case. Normally, most users, most of times, don’t care about the specific keyboard is connected to the zynthian, and you don’t want to see warnings every time you load a snapshot that was created with a different keyboard or controller connected.

I’m pretty sure you understand we have to think in general cases before specific cases, and majority’s cases before single user cases. This is how this works. Of course, if we had several users complaining about this, we could find a way to implement the hardware checks without annoying the rest of users, but AFAIK, you are the only one complaining about this, and we don’t have the resources to implement the specific use-case of each specific user.

Anyway, and as @pau already recommended, you should take care of testing your setup before start playing. AFAIK, most of musicians do it :wink: