any news on further development on the Soundplane software, in particular the touch tracker? are you likely to be looking over that code base any time soon?
Ive been doing some more work on MEC, my project which provides a standalone solution for the Soundplane (and Eigenharp), basically turning things like rPI/BBB into 'intelligent dongles' , so the Soundplane can be more like 'a standalone instrument' rather than a computer controller.
https://github.com/TheTechnobear/MEC
currently works with Eigenharps and Soundplane, and connecting too MPE devices (im using the Axoloti to make it computer independent :))
if your back into 'Soundplane' mode, it'd be great to discuss a few thoughts i have. as Im getting a little bit too much latency with the Soundplane due to the CPU requirements of the touch tracker.
so not sure if I can get some performance gains, or really need to move to something more powerful, the O-Driod C2 is quite a tempting candidate, given the rPI3 wont work.
anyway, I know, not much point in discussing unless your actively looking at the Soundplane software... since I think its quite a while since you were look at that code base.
The Soundplane software will get another release ASAP. First I have an issue with Kyma support to take care of. This should be quick. Then I can work on the touch detector again. The new detector should be significantly less heavy in CPU than the old one. Efficiency was not a big concern at all for the initial Soundplane release, so i'm confident I can do better.
I have to get a bugfix release of the instruments out first. At that point I'll be back to the Soundplane full time.
I’m glad that the MEC project is on the way. Thank you a lot for this work. I also want to get out of the desktop ou laptop computers. They aren’t so efficient for live music, and if I play this beautiful wood instrument, it’s not for having a big screen in front of me on stage.
For this purpose, I own a Kyma system, an axoloti and a beagleboard bela. But I'm a beginner with these systems (more on Max or Pd) and it’s really difficult to have something that work well with the soundplane. The connexion with Kyma is really buggy. I hope that the work that you are doing Randy is going to improve things.
Let me also propose you some other improvements on the soundplane software for live performances :
Sometimes touches get stuck (but mostly at the beginning, it doesn’t happened from a long time for me, maybe the sensor is better after playing a lot). For having a safe live system, it will be really great to have a possibility to send a midi message (for exemple with a pedal) to flush the system in case of a problem.
When we play a really long time or when we reach the limit of the CPU (which is easy with aalto or kaivo), an ultra big latency appears. I think that it’s because OSC isn’t on the high priorities of a normal OS (another reason to quit "office" computers for live music).
This is really problematic. Again, is there a way to send a message to init this, without having to quit the client in the middle of a live performance ?
The edges of the soundplane aren’t well detected. At the top and the bottom, the row begin at the middle of the square. I can correct it on max or on kyma, but a possibility to correct this directly on the client will be great (especially for Aalto and Kaivo).
It’s impossible to use multiple OSC sends or to dynamically change OSC send address on the client via midi. That made the limitation to only use one synth at a time (and one synth per performance if you doesn’t want to touche the computer on stage).
Before, I was using Max for dispatching OSC, but it's CPU heavy and the multiplication of the softwares made things less stable.
Idem for the zone preset. Before a performance, I have to choose wich zone I want and I cannot change it by midi program changes. It’s always a dilemma between continuous surface or scaled one (especially for the Y), because Aalto or Kaivo cannot put this directly in their presets (I change them on stage with a pedal).
Last thing : if you cannot succeed to improve the touch detection, I suggest you to change the grid layout on the B model for matching the detection. It’s a little bit frustrating to be unable to hit two adjacent squares and to havent any landmark that matches the reality of the touch tracker.
Thanks
Thanks for the detailed feedback. I'm getting back to the Soundplane software now.
The first thing I'll be working on is the new touch detector. Hopefully this will improve CPU usage a lot, as well as the edge effects you are noticing.
A lot of these other things were on my list too, and now they all are.
I would like to improve Kyma support, but it's a bit slow for me to get feedback. Maybe I could get a loaner machine from Symbolic? I'll ask them after I can get some of these more major things under way.
Great. Thank you Randy.
I have some great news for Soundplane owners :)
the Raspberry PI kernel bug (in dwc_otg) which prevented the SP working has now been found/fixed. ( not released yet)
Yesterday, I ran a patched kernel on a rPI2, and have the SP working perfectly using MEC!
my setup:
Soundplane + rPI2 + Axoloti (midi mpe)
all perfect, Soundplane using 80% cpu on 1 core, other 3 cores very low cpu.
(Eigenharp tested too, and only uses 10%, others low)
the Soundplane and axoloti directly powered from the rPI2, so connect this to a USB battery and you have a completely portable setup, just add headphones/speakers :)
the rPI2 probably has enough cpu power left also to run a (light) synth... or you can add another couple of Axolotis or Belas (you have 4 usb ports to play with)
p.s. with MEC on a PI (or anything else), you can also send T3D OSC messages over ethernet or wifi ;)
use a rPI3, or an Asus Tinkerboard (mine arrived a couple of days back) , and you have even more cpu to spare.
of course it will still be great for the TT cpu to drop, as then the BeagleBone Black with Bela could be used, Bela providing low latency audio, and analog out (useful for modular cv ;))
oh... I've also done a successful technical test with the Eigenharps using libusb running on windows. (with usb iso traffic).
this means when my windows laptop arrives, I will be able to get MEC running on windows, and yes that includes the Soundplane on windows!
Im really excited, the fixing of the PI2/3 make this so much easier for everyone,
its freely available and only $35 (ok, perhaps tiny bit more for case, power, sdcard)
once Ive 'finished' mec, Im going to look at using buildroot, to turn this into an embedded appliance...so users just see the MEC/PI as a 'magic box' you just plug in.
Great news! This kernel bug was preventing isoch USB, is that right?
I'm excited to see your setup in action when you have time to document it. I would also like to put together a guide to making the raspberry Pi solution work and get it up on the web site.
This also fires me up to complete v.2 of the touch detector.
Are you using the Linux Soundplane driver that Per Eckerdal came up with? That was created "on the way" to getting a Windows driver working, but we never got all the way there. If you need any kind of assistance or encouragement with the Windows work, let me know!
MEC seems like really interesting work and I can't wait to check it out.
Waou, fantastic. Can't wait to test this in my setup ;)
Hi randy, How is the new touch tracking software going?
I've seen the check-ins on the repo, is it a stage where its worth playing with? testing? have you been mainly working on detection, or also lowering cpu loads?
If there is anything I can do to help, let me know
Cheers
Mark
p.s. great to see you having some time for this, thank you for your efforts.
Working on both-- the new detection takes a lot less expensive filtering. I'm excited about it. I'll be at beta within a week, I'm pretty sure. I'll clean it up for that release, so I'd wait just a bit. It will be very helpful if you can try compiling it for Linux at that point!
excellent work :)
ok, I'll hold off a bit, shout when your ready - excited here too!
I'm having trouble finding a link to download the soundplane software... I finally bought a music MacbookPro
Is in the first sticky thread, on this sub forum ( Soundplane client for Mac)
http://madronalabs.com/topics/1691-sticky-soundplane-client-for-mac
Probably should be a link to it in the Soundplane production page too.
(Or perhaps I missed it ?)
Sorry it's a bit hard to find now. I'll work on a documentation pass after this Soundplane beta coming very soon.
oooh, beta... now I'm excited :)
I'm pretty sure this is the longest I've worked on any single technical problem, aside from my thesis work. I think it's possible to do a great touch detector with the Soundplane A sensor, but, just barely. It will be a big relief to roll out this release and get your feedback.
Sounds like an excellent challenge, look forward to seeing the results