Sure, let’s create a dtbo overlay, past the content of uart-speed script, then save it somewhere, sudo cp it into /boot/firmware/overlays. Nano into config.txt
and add dtoverlay=uart-speed.dtbo. reboot.
I have no idea what that does. Vcgencmd measure_clock uart hasn’t changed.
It is still 48001464
Despite. I am using a Pi4B with Bookwork 64 bit, with latest rpi-update.
I have already installed ttymidi-sysex and bento-ttymidi. I gave up on Pi5B some time ago, I can use Usb Midi there. ldrolez-ttymidi has only been tested up to Pi4, so
because he’s also the maker of my Midi hat, I have been trying to follow his instructions. And they aren’t working. So now he suggested me to open an issue on his github.
I also tried install python -m venv and install ttymidi over there, I’m still searching
for mod-ttymidi as used in Zynthian. There is zero documentation on the rpi forum.
Even if venv should only be useful for Pi5 models, I had problems with the Pi4 too
when trying to install WiringPi, so I installed it there.
The thread you linked was locked, So any kind of feedback was cut off.
Now I have opened Yet another thread, release the MIDI baud on pins 14-15.
I can monitor the MIDI input with Gtkterm, set tty0 port to 38400 for listening, but it’s dead silent. My SQ64 is connected to Midi in of the SLIM MIDI HAT, playing a random pattern constantly.
Whatever I try to do, I can’t listen to an external Midi.
The problem is there are issues evidently, and not enough testing.
Because I have been testing non-stop for two months now…
There is not one seamless guide to follow, it’s all a patchwork
of fixes after another. This is my problem.
I have a Midi transposer running on an Arduino Nano, and a Raspberry Pi Pico.
It’s a simple script, Midi in, Midi out. It just works. Easy.
On the Pi4-Pi5 it used to be fairly easy untill 2018, then someone
started having problems. To me it seems that whatever caused the first
time the old methods to go belly up, it’s still there, lurking beneath.
Whatever issue seems to be approached with a “fix” but there is no
time to wait for someone to actually test the various scenarios and
Pi4, Bullseye, Buster, Pi5, Bookworm, everything is merged together
and it’s impossible to get a clear picture of anything.
Only now, after months, it finally seems that someone is trying to get to the core
of the issue. But we are far from over.
The Cherry on top was “but it seems that on Zynthian the issues has been resolved”…
https://forums.raspberrypi.com/viewtopic.php?t=389528
PhilE wrote: ↑
Wed Jul 02, 2025 3:26 pm
So you either try to set a custom baudrate using the UPF_SPD_CUST feature to get 31250 baud or you use the overlay and ask for 38400 baud. With the overlay, that whole set_custom_baudrate function in the bento_ttymidi bridge is unnecessary.
I confirm! The bento code seems to be buggy since it tries to set the baudrate to 38400 and then 31250. Moreover the code does not handle Sysex properly.