It uses an 8x8 LED matrix with the MAX7219.
Now I have a fairly high continuous tone (1 - 2Khz) in the audio output, which changes slightly with the content of the LED display. When I turn down the display brightness in the code, the continuous beep gets quieter, but doesn't go away.
I have already searched the forum for similar posts, but somehow there was no solution or the solution was not really revealed.
So I'm trying my luck with a new thread.
I have attached the schematic and two photos of the synth for you. I hope this helps to narrow down the problem.
I've tried a few things. e.g. 100nF and 10uF capacitor at +5V and GND of the MAX7219 module.
By coincidence, I was able to achieve quite good damping of the continuous tone if I connect pin D9 (audio out) and pin D10 (NO button to GND) with a 100nF capacitor. But I can't explain it.
The scan frequency of the matrix dispaly is somewhere between 500Hz and 1300Hz, so as you already know, that is where the tone is comming from.
You are taking power and ground for your audio circuits from the matrix display board. (outlined in red). That may be how the interfering tone is being coupled in.
Disconnect those wires from the matrix display and connect them directly to the +5 and ground with twisted wires, like your other connections.
Thanks for the tip about separating the power supply from the LED display and the analog part. I implemented it directly as you recommended, but unfortunately it didn't bring any improvement.
What I also noticed, at the beginning the device shows a bright flashing logo. During this time, you still don't hear anything from the background noise. Only when the PWM of the Mozzi library starts does the continuous tone begin.
Attached is a photo of the new wiring and a short video (sorry have to zip it) that clearly shows the start process.
Do you have any other suggestions?
Your original diagnosis definitely made it sound like the display was the culprit but now I don't know.
Maybe someone with experience with the Mozzi lib can help.
What happens if you completely disconnect the display?
You might need a separate voltage regulator for the LED section. Voltage regulators make pretty good filters if you can keep the noise on the high-voltage side.
Or if the 9V is a battery, maybe also a separate battery.
As you probably know, PWM isn't analog and it can be noisy or it can do "bad things" to an audio amplifier if not "completely" filtered.
And just mixing analog and digital can be tricky... It's done "everyday" but the digital signals have to kept-out of the analog because our ears are very sensitive to noise.
Many thanks for your help.
I searched the forum and the internet a bit.
I came across this post by Grumpy_Mike. post
Then I found this video about the MAX7219 on Youtube. YT video
I then combined these two pieces of information and designed a small filter and pull-down circuit for the MAX7219. In addition, I soldered a 100nF capacitor directly to the MAX7219. That worked very well, but the 1000µF capacitor had the greatest effect.
The whistling of the MAX7219 is now almost inaudible and the display now runs much cleaner when powered up.
Attached are two photos of the LED module and the filter circuit as well as the schematic.
Maybe this will help others with problems with the MAX7219.
I implemented it directly as you recommended, but unfortunately it didn't bring any improvement. 
