Yoshimi and Poly Aftertouch

[Folderol]

For some years now, we've had poly (key mode) aftertouch in Yoshimi (classed as experimental), but I'm not rich enough to be able to buy a keyboard that sends this, so have had to rely on faking it for testing. I think it's right but would like to know for sure.

If anyone has such a keyboard, can you make the following test please, to see if it is in fact correct.

Set up a fairly harmonically rich part, and in the aftertouch window set just poly (Key) aftertouch for Filter Cutoff.

Then, lightly hold a chord on the lowest available keys of the keyboard, then play a single note high on the keyboard, changing pressure on this should activate the Cuttoff change for just the high key without disturbing the held chord.

Do the same for Filter Q and Pitch Bend - also combinations of all three.

Modulation will require an instrument that has significant frequency LFO depth.

Bear in mind that all of these are not absolute settings but relative to the value set by default values as modified by incoming CCs.

If you can produce a recording of the results that would be fantastic.

[ajay_m]

I will try and see if I can do this for you maybe over the weekend and let you know, I have both the hydrasynth and the qunexus, both of which do poly after touch. I assume, never having used yoshimi, there's a pc version or do I have to break out one of my stash of raspberry pi's (I know, possessive apostrophe not required, just looks a bit rude without

[James Perrett]

Yoshimi is currently Linux only but works well on a Raspberry Pi.

[Folderol]

Thanks ajay. I don't know which version of Yoshimi is currently supplied on RaspberryPi OS, but it usually works perfectly well on Pi 3B or later. You will always get better results if you self-build Yoshimi though.

As for other platforms. We've twice been approached by MacOS people. They've looked at the code, then we've heard nothing more
This is a bit frustrating, as we don't know why. It is extremely complex software. So is it due to that? Or do they think it's badly written?
We've not had anything at all from the Windows people, and I expect it would be rather difficult to maintain. Microsoft seems to chop and change all the time.

[James Perrett]

We've not had anything at all from the Windows people, and I expect it would be rather difficult to maintain. Microsoft seems to chop and change all the time.

Not sure where you get that idea from Will. I have plenty of 20+ year old Windows software still running happily on modern computers. And the most popular virtual synth format (VSTi) is nothing to do with Microsoft (though Steinberg are possibly worse when it comes to preserving compatibility over the long term). Windows users can also have a play with ZynAddSubFX

https://zynaddsubfx.sourceforge.io/index.html

which is what Yoshimi was originally based on as I understand it.

Plus, if you want something cross platform then it might be worth investigating the CLAP standard.

[Folderol]

We've not had anything at all from the Windows people, and I expect it would be rather difficult to maintain. Microsoft seems to chop and change all the time.

Not sure where you get that idea from Will. I have plenty of 20+ year old Windows software still running happily on modern computers. And the most popular virtual synth format (VSTi) is nothing to do with Microsoft (though Steinberg are possibly worse when it comes to preserving compatibility over the long term). Windows users can also have a play with ZynAddSubFX

https://zynaddsubfx.sourceforge.io/index.html

which is what Yoshimi was originally based on as I understand it.

Plus, if you want something cross platform then it might be worth investigating the CLAP standard.

You managed to hit a number of sore points there.

The last version of Windows I actually used was 95, so I'm quite out of touch with it. At the same time, I'm seeing a continuous stream of complaints about it pretty much across the board. I don't think I want to go there.

I'm quite well aware of CLAP... and all the other 'standards' that want you to rewrite all the communications (and display interface in some cases). With a lot of these, your project loses a significant part of the identity that differentiates it from all the others.

However, that's not quite so bad if you have a huge team of professional coders - Yoshimi has two experienced amateurs, one very busy professional - when he has the time, plus a few who drop in on odd occasions. I'm the only one who's anywhere near full time.

I won't get into 'my synth is better than your synth' type arguments, but will point out that Zyn and Yoshi these days are very different beasts.

The good news?
We must be doing something right - we have a very loyal band of users and helpers.

[James Perrett]

You managed to hit a number of sore points there.

Apologies Will. I hear lots of great sounds coming from my son's room when he's using Yoshimi so it is somewhat frustrating that it isn't cross platform. Unfortunately my main audio interface isn't Linux compatible so I can't switch OS.

[Folderol]

You managed to hit a number of sore points there.

Apologies Will. I hear lots of great sounds coming from my son's room when he's using Yoshimi so it is somewhat frustrating that it isn't cross platform. Unfortunately my main audio interface isn't Linux compatible so I can't switch OS.

And I'm sorry for that outburst!
It's frustrating for me too. I'd love to see Yoshimi on a broader base, but for that we need people to join us that really know the platforms and are willing to spend time getting to know our code... warts and all.

It should be easier than it looks on the surface. Jack is cross platform and so is FLTK - our GUI library. The entire filesystem works though a single header which shouldn't be too hard to replace. For the other systems ALSA audio could be replaced with PortAudio. However I have no idea how MIDI is handled elsewhere (although Jack MIDI should be able to work).

Input using the command line should be pretty similar for the Mac. I've no idea what would be needed for Windows, so that might have to be abandoned.

[James Perrett]

From a user's point of view it would have to be a plug-in so it is up to the host to handle the raw Midi and audio routing. It would also be a good idea to use a cross platform framework like Raylib. You'd be amazed at how quickly my lad can code impressive looking software on Raylib which can easily run on both Linux and Windows (he currently doesn't know anyone with a Mac but it would probably be just as easy to port to Mac too).

His current project is intended to eventually be something like this

https://www.bespokesynth.com/

although currently it is much more limited.

[BJG145]

Yoshimi and Z*** have a troubled past, so I thought Folderol was quite restrained, but we won't dwell on that. I rarely use Linux, but I'm interested in RPi and I've been daydreaming about a headless Yoshimi on that platform for years...the thing can sound amazing, and I'm gradually working through options for embedded audio. (Currently on Ksoloti.)

Also noticed a VST wrapper for LV2 plugins on Github which might be worth a look? But I'm not familiar with LV2.

[James Perrett]

Yoshimi and Z*** have a troubled past, so I thought Folderol was quite restrained, but we won't dwell on that. I rarely use Linux, but I'm interested in RPi and I've been daydreaming about a headless Yoshimi on that platform for years...the thing can sound amazing, and I'm gradually working through options for embedded audio. (Currently on Ksoloti.)

I hadn't followed that history - I only really became interested in Yoshimi when my lad acquired his first RPi courtesy of another forum member here. I would have thought that you wouldn't need anything other than the RPi, a DAC board and a Midi in/out board. Something like Ksoloti will probably become outdated fairly quickly whereas a straight RPi is more likely to preserve compatibility between different versions.

[Wonks]

Ksoloti?

Bless you and get better soon.

[BJG145]

I won't derail Folderol's post, but Ksoloti is interesting. Not as powerful as RPi/Yoshimi, but it's apples and oranges. It's recently been brought back from the dead (it dates from 2014), but I'll post more on that elsewhere.

[Folderol]

One thing that I'll never forget, was when a number of Linux audio people were vociferously demanding that Yoshimi should be abandoned when Jeremy (who had quickly stepped in when we lost Cal) said he could no longer work on it. It was probably that, along with one person privately encouraging me to take over the reins, that was the ultimate decider. Quite few others kept their heads down until I actually took control, then (again privately) offered support.

[The Elf]

Folderol knows how often I've marvelled at the beautiful sounds I've heard him conjure from Yoshimi. He also knows that I've often bemoaned that Yoshimi isn't a (Windows) VST plug-in. If it were I would have adopted it long ago.

But I understand how complex modern code is. It would need some serious effort, way beyond the typical tinkering reserved for marginal time.

Folderol - your boundless support, enthusiasm and promoting of Yoshimi does you credit. You're the best advert it could have. I remain a non-user, but I'm still cheering from the sidelines!

[Martin Walker]

Folderol - your boundless support, enthusiasm and promoting of Yoshimi does you credit. You're the best advert it could have. I remain a non-user, but I'm still cheering from the sidelines!

Hear hear!

[ajay_m]

Yes, this is the true internet spirit. I've never played with yoshimi but my curiosity is piqued and I have an rpi4 with 2G twiddling it's little electronic thumbs, and my fresh shipment of 1.3 inch oled displays arrived a week ago ( used the last one to build a train departure board for a grandson who is crazy about model trains). Now I have to fit this in with the ongoing hp vintage calculator restoration projects etc but some kind of little sound module thingy could be interesting as a project.

[James Perrett]

He also knows that I've often bemoaned that Yoshimi isn't a (Windows) VST plug-in. If it were I would have adopted it long ago.

My lad reckons you can get it running under Windows but it involves something like this

https://learn.microsoft.com/en-us/windows/wsl/install

and I don't think that he has actually done it himself yet.

However, this morning he commented "Why does Yoshimi never take more than 2% of my processing power no matter how many notes I play?" which is a testament to Will's code.

[BJG145]

Interesting...I previously got Yoshimi running on Windows via VirtualBox, but I hadn't come across WSL before. Latency was a problem with the VirtualBox setup, and I'm wondering if this might also be complicated to address via WSL. Tempted to have a go, but given the recent post by Eddy on the performance under RPi5 I think I'm even more tempted to grab one of those.

I booted up my 4GB RPi 3B+ yesterday, but being used to a 16GB i5/SSD I was a bit frustrated by the speed of the GUI. I'll try an 8GB RPi 5 though. Y'all have talked me into it.

(I'm especially interested in finding something I can manage from a DIY controller with no QWERTY keyboard or HDMI video, as I can do with microcontroller-based embedded audio options like Daisy Seed, so I'm curious to find out whether that's possible. Korg use RPi compute cards in some of their synths, but I expect their bare-metal programming skills are some way ahead of mine.)

[BJG145]

PS Here are examples of products built with Daisy and Ksoloti respectively. This is the sphere I'm interested in at the mo. I'm wondering if I can use RPi for things like this.

https://soundwork.shop/wiggler-1
https://www.artiuminstruments.com/the-swarm

Organelle is a comparable device built with RPi. I'm looking for a way in.

https://www.critterandguitari.com/organelle

[BJG145]

RPi 5 on order. Eddy/Folderol, stand by for Yoshimi questions...

[Folderol]

He also knows that I've often bemoaned that Yoshimi isn't a (Windows) VST plug-in. If it were I would have adopted it long ago.

My lad reckons you can get it running under Windows but it involves something like this

https://learn.microsoft.com/en-us/windows/wsl/install

and I don't think that he has actually done it himself yet.

However, this morning he commented "Why does Yoshimi never take more than 2% of my processing power no matter how many notes I play?" which is a testament to Will's code.

Hmmm. Not sure what's going on there, or what he's actually measuring.

I find that Yoshimi typically takes 2% when Idle, about 5% when playing a handful of notes with mid range instruments and 20-30% when working hard.
Also this is deceptive, as all measurement systems I've seen take partly averaged occasional samples. You only know if you've pegged it if you see Xruns. Even that's not conclusive as Xruns happen when the total load from all programs hit the buffers.

Finally, much though I appreciate the thought, it's not my code. I'm probably responsible for about 30% of it. The 10% of critical core code is above my pay grade

[James Perrett]

PS Here are examples of products built with Daisy and Ksoloti respectively. This is the sphere I'm interested in at the mo. I'm wondering if I can use RPi for things like this.

You can definitely use the RPi for things like that - what you can do is only limited by your imagination and your ingenuity with I2C or similar control protocols.

My lad has gone the other way. His RPi based synth is influenced by the controls on the early Yamaha digital keyboards which used a small 16x2 character display and the minimum of knobs or buttons. He uses one display and one rotary controller to control everything with his own custom synth software. It is deliberately difficult to use!

[The Elf]

He also knows that I've often bemoaned that Yoshimi isn't a (Windows) VST plug-in. If it were I would have adopted it long ago.

My lad reckons you can get it running under Windows but it involves something like this...

But that would still leave it outside my DAW.

[James Perrett]

My lad reckons you can get it running under Windows but it involves something like this...

But that would still leave it outside my DAW.

Yes, you'd have to run virtual Midi cables and find some way to route the audio.