MIDI Completely Non-functional - Akai MPK25 + Win 11

[craigvt]

Setup is windows 11 on current updates, with an AKAI MPK25 midi controller, using the software in standalone mode.

The overall issue appears to be that when any midi event is triggered from the controller, every receiver in the application responds to the input. It happens regardless of a note or cc command. For example, if I press pad A1, every function on the app responds, every pad lights up, the play starts, 16 levels is activated, the jogger turns, record starts, and the sample edit mode launches. The best way I can describe this is that some kind of issue in the midi mapping is sending the command everywhere instead of where it is supposed to go.

I've tried everything I can to try and eliminate user error. Including;

• New midi preset with everything to CH ALL
• New midi preset with everything to CH 1
• Removing all velocity, after touch and other things from the controller so its only sending basic note/cc data
• Using a completely blank slate preset on the midi controller
• Everything set to OFF except a single control
• Changing every possible setting in the midi/audio window
• Tested on laptop and desktop
• Tested in Studio one, Ardour and MPC desktop as a VST

I should also mention that within the settings>midi window is the only area of the application I've experienced severe crashing issues;

• Using the save function leads to frequent crashes ( using open window > save > replace does not do this)
• Sending any type of midi data frequently results in a crash ( likely due to the above )
• Trying to name a midi preset does not work and will lead to a crash at times. When moving to the next character, the previous character is replaced by a random character, resulting in the inability to actually name a preset.

I'm hopeful this is just an issue with controller or something I'm doing wrong. I'm probably going to test it out on a linux machine because I'm super curious if this is a Windows problem.

Please let me know if I can provide any other info. Thank you for the work you've done on this project!

[izzyreal]

Many thanks for the detailed report. I'm quite sure there's a bug in the MIDI mapping and the persistence of MIDI mapping presets, and it's probably not platform-dependent. I'll try to reproduce it and investigate in the near future. It would be helpful if you provide files you have in ~/Documents/VMPC2000XL/MidiControlPresets, as well as midicontrolmapping.vmp from the config dir (see https://vmpcdocs.izmar.nl/0.9.0.0/install.html#file-locations).

[izzyreal]

I've decided to retire the custom binary format of the MIDI presets and switch to JSON, supported by a schema definition. Currently WIP here: https://github.com/izzyreal/mpc/tree/midi-presets

[craigvt]

Thank you for looking in to this!

Here are the requested files. The MPK preset is the one I've created. I did manage to get the pads working. As you can see in the file, I've turned everything to OFF with the exception of a few transport controls for testing. Even with everything set to off, most functions still respond to MIDI input.

vmpc-files.zip

[izzyreal]

Still work in progress. I ended up rewriting a huge chunk of https://github.com/izzyreal/mpc. There were many issues with many things, including the hardware model and input routing and handling. Both are crucial for your issue.

Especially the input handling is a tough nut to crack, and the project had accumulated a lot of different ideas over the 10+ years that I've worked on it, and none of them really worked well. The result was a very messy and incoherent codebase, and therefor a bugprone project (as you have experienced firsthand).

In any case, now I have something that I think should be better. It's likely I've introduced regressions and new bugs, but the architecture is hopefully conducive to fixing these things relatively easily. I've also worked on a new file format for MIDI control presets as mentioned, based on a JSON schema.

I'm nearly done implementing the original 2000XL MIDI input features. After that I want to take another stab at what custom, non-original MIDI control should look like exactly for VMPC2000XL, and integrate the new MIDI control preset format. I'll let you know when I have something worth testing. I hope it will be worth the wait!

[craigvt]

I appreciate the follow up and having a look to see if it's something that can be implemented. You make a good point about the full midi input being a custom distinction from the original and perhaps that type of setup is out of scope. Essentially, what I was trying to achieve is having this on my laptop and turning my controller into an MPC, with no keyboard or mouse, just fully controlling the software with the controller. That's where I ran into problems, for example setting the shift button to a pad, which sends note data instead of CC, ect ect. It would work awesome until I tried to save it lol.

It's definitely worth the wait in my opinion. I've looked for a project like this forever. Always blew my mind that more full hardware emulation of classic samplers is not done. The RX1200 is a great project, but they didn't include the sequencer or sampling, which kind of just kills the stand alone viability. Thanks again and take care!

[izzyreal]

You make a good point about the full midi input being a custom distinction from the original and perhaps that type of setup is out of scope. Essentially, what I was trying to achieve is having this on my laptop and turning my controller into an MPC, with no keyboard or mouse, just fully controlling the software with the controller.

To be sure, what you want is most certainly not out of scope. It's exactly one of the features that I want to see work smoothly with VMPC2000XL. But architecturally it's a challenge, because as an emulator it should as a minimum do what the original does, as well as offer different and convenient ways to control the emulator via MIDI. Some motivating examples in this context:

1. The original MPC2000XL allows simulation of physical pad presses via the MIDI footswitch assignments. The caveats here are that it only allows to simulate pad presses at full level, and MIDI footswitch assignments only work with CC messages. For VMPC2000XL, however, as a user I would expect to be able to simulate physical pad presses via MIDI note on messages as well, and to be able to make use of the velocity of note on events to translate to pad pressure.

2. The original MPC2000XL allows simulation of certain button presses via the MIDI footswitch assignments. Here too, the caveat is that it's CC messages only. For VMPC2000XL, I want users to be able to trigger button presses via note on messages, and releases via note offs.

3. Although the original MPC2000XL allows simulation of certain button presses via MIDI, some buttons are excluded, such as OPEN WINDOW and MAIN SCREEN. There's also no way to turn the DATA wheel via MIDI, or the REC GAIN and MAIN VOLUME knobs. With VMPC2000XL basically anything is in principle controllable in any kind of way, if it is implemented as such. And that's what I'm working on right now.

[izzyreal]

Sorry for the long wait. As mentioned, I ended up rewriting much of the code base. I just ran some statistics, and it's fair to say I rewrote the whole thing, which is over 100k lines of code spread over 1000+ files.

In any case, if you still want to use VMPC2000XL, I have built a preview for the upcoming release so you can test the new extended MIDI controls: https://drive.google.com/file/d/1aD_WxcYuWmsnksXDJAnBUYSLLLq77Ltw/view?usp=drive_link

I haven't updated the documentation, but the GUI for extended MIDI controls has remained largely the same. Quick summary of new stuff:

• Button-like targets (i.e. all actual buttons and DATA wheel - and DATA wheel +) have a v: column. This value serves as a threshold, meaning CC events with values equal or greater than the value are interpreted as "press", and below the value as "release" (where a "release" doesn't do anything for the 2 DATA wheel targets).

• The singular data wheel target (without - or + suffix) is meant to be associated with an endless dial on one's MIDI controller. This new build of VMPC2000XL anticipates 2 common encoding schemes, which are named RSL (relative stateless) and RSF (relative stateful). The Akai MPD218 that I have emits RSL when in the MPD218 editor I set the TYPE: for a given knob to INC/DEC 2. The MIDI events for this encoding scheme correspond with CC events with values between 1 and 63 for clockwise turns, and values between 64 and 127 for counterclockwise turns. Assuming your Akai controller supports a similar encoding scheme, I won't bore you with the details surrounding RSL.

Let me know how it goes. I'd love to make everything work well with the MPK25!

[izzyreal]

In the meantime the changes have been made part of the official release: https://izmar.nl/release-candidate