would be feasible to use a raspberry pi 1B? i just want to use munt with MT32 emulation only

Isn't 1B single core 700MHz CPU? That would probably not be workable..

I got a quick question regarding the Wavetable header. I tried using the MIDI from the header but I can't seem to get it to work. Is MIDI simultaneously output on the Wavetable header and the game port or is a config needed for that?

weldum wrote:

would be feasible to use a raspberry pi 1B? i just want to use munt with MT32 emulation only

So far we've only been able to get Munt working reliably on RasPi 3 and above. Only soundfonts will work with a 1B.

ElBartoME wrote:

I got a quick question regarding the Wavetable header. I tried using the MIDI from the header but I can't seem to get it to work. Is MIDI simultaneously output on the Wavetable header and the game port or is a config needed for that?

It might depend on the soundcard, but it should be directly tied to the same line that feeds midi to the gameport. If mpu401 emulation is enabled in the soundcard both the gameport and the wavetable header should be working.

I have a ESS1868 soundcard. Unfortunately I can't get it to work and changed back to the game port for now.

I finally updated the 3D model of the cover. The new LCD is a lot bigger than the old one so it takes almost all of the space.
I also added a mount for the Pi. It just snaps into place without screws. The 3D model already has supports for the mount implemented. After printing you have to remove the support.

Here some pictures of the new cover: https://imgur.com/a/FS7wmBQ

Next step is to modify the code...

That's perfect, nice job!

t9999clint wrote:

weldum wrote:

would be feasible to use a raspberry pi 1B? i just want to use munt with MT32 emulation only

So far we've only been able to get Munt working reliably on RasPi 3 and above. Only soundfonts will work with a 1B.

so no luck for me 😢

No, you need a faster CPU than that to run Munt.
I would suggest getting the cheapest rpi4 available if all you want is mt32

I found a few projects like this one here https://github.com/MarquisdeGeek/FluidPi
that I could use to implement a webserver to control WavePi, but it uses a Bashscript as a webserver. Every security conscious part of my mind is telling me that this is a bad idea, but other than the security concerns it actually seems like it'd work pretty well. The code is super light and portable which I think fits the overall theme of the WavePi project perfectly. I kind of need a second opinion on this...

Should I implement a simple bashpowered http server just for controlling the WavePi? Or should I try to implement something else on a more hardened but much heavier system, like LAMP or something...

What did the PiCorePlayer guy use for his WebUI?

gdjacobs wrote:

What did the PiCorePlayer guy use for his WebUI?

Kind of looks like a hodgepodge of a couple of different technologies. I don't think I want to implement something like that.

My original plan was a basic PHP powered site that just gives a list of profiles and soundfonts Ave let's you switch between them. The FluidPi project has some code that allows for this without even needing PHP or apache installed. It looks like an elegant solution, just not really good security wise.

I guess I can try to lock it down a bit, a non standard http port and a authentication system to prevent random attacks. I was just wondering if anyone with a security background could drop some knowledge on us.

Falcosoft wrote:

Neon/vfpv4 can help in case of floating point rendering (just like SSE2 on x86) but using floating point rendering mode of Munt is not realistic at all on Raspberry Pi 2. It can even kill 2GHz+ x86 CPU's in some cases.

I suspect what's happening here is the code generator activating the Neon unit for integer instructions as well, probably because of -mcpu being specified.

On the Pi 2 issue, I did a smoke test using my trusty Loom midi dump on a brand new Devuan 2.0 image. Everything worked okay, although the Pi 2 is definitely at it's limits. So, it's viable, but that recommendation might have to be revisited if Serge implements further changes requiring additional IPC.

I did have to make sure cpufrequtils was installed. Without it, the A7 cores stayed at 600mhz and Munt was starved for cycles. I've added a note in the OP of my Munt on Pi 2 thread regarding this.

gdjacobs wrote:

I did have to make sure cpufrequtils was installed. Without it, the A7 cores stayed at 600mhz and Munt was starved for cycles. I've added a note in the OP of my Munt on Pi 2 thread regarding this.

Ok, so in theory all that's left to do is add cpufrequtils to the dependencies in order to fix the performance issues with raspberry pi 2 boards?
Are there any commands that I need to run with cpufrequtils in order to prevent underclocking, or is it automatic?

The current CCFLAGS and CXXFLAGS settings for the raspi 2 are "-Ofast -mcpu=cortex-a7 -mfpu=neon-vfpv4 -mfloat-abi=hard -mvectorize-with-neon-quad -ftree-vectorize -funsafe-math-optimizations"
(from Line 69 of: https://github.com/t9999clint/WavePi/blob/master/install.sh)

Check my thread. I added the diagnostic checks to see if the CPU is being held at the base clock or if it's ramping with process load. There should be an init script that loads a clock governor profile to the kernel. Hopefully TC includes that as part of the package.

I think Zoltán (Falcosoft) is correct in that we don't have to worry about FP optimization very much. I suspect we're fine as long as we use heavy optimization flags (-O3 or -Ofast) and an appropriate -mcpu/-march + -mtune combination to pipe integer calculations through the Neon unit. Still, I'm not going to try fixing what ain't broken.

Incidentally, -Ofast implies both -ftree-vectorize and -funsafe-math-optimizations, so you don't need those flags included. I'm not sure if the GCC code generator emits both quad and double vectors by default now, so that flag might have to remain. Side by side testing indicates the the effect is negligible if anything, so with newer GCC versions it might not have a benefit anymore.

Released a new version of my KOR Soundfont. This one I made specifically for the WavePi.
It's only 600mb!
https://downloads.kor.ninja/Music/korfontproject/?dir=tiny

If you don't mind the extra loading time, this is what I'd reccomend to use on a RasPi 3 or 4 (with 1gb ram or higher)

Hi everybody,

I'm having some issues with the serial connection on my wavepi.
some info:

- I have wavepi running on a RPI4 with the 20x2 LCD display.
- The serial console is turned off.
- I'm using a clone MPU-401 card for the MIDI output.
- Everything works fine when I'm using the roland UM-ONE usb-midi device connected to the wavepi, midi messages appear on the LCD display and sound works.
- I have build a circuit based on this one: "http://www.smbaker.com/wordpress/wp-content/u … matic-orig2.jpg" (ignore the 5v rail, everything works on 3.3v)
- I have built the "MIDI IN" and "MIDI THRU" ports.
- When I connect the roland UM-ONE (connected to the RPI) to the "MIDI THRU" port everything works fine so I know the circuit works because the "MIDI THRU" port takes the same output from the 6N138 as the serial port.

Ok, so now the problem: (I don't use the UM-ONE from this point on)

When I connect the output from the circuit to the RPI (MIDI_RX on PIN 10) and play Space Quest 3, I don't get any messages on the LCD display. The Wavepi plays random notes instead of showing the message on the screen. It tries to play the message instead of showing it on the screen.

So i tried installing the latest version of Wavepi but now I don't get anything, no music, no random notes, nothing. I do however get a lot of messages on the monitor that is connected to the HDMI port (yes, serial console is turned off).
When I use, TTYmidi -s /dev/ttyAMA0 -p (print command), it prints the number "80" every second and a lot of other numbers when I play some midi music. Soo it does receive data from the MIDI circuit over the serial port.

I hope my explanation is clear enough.

Thanks in advance.

For clarification, and please correct me in case I made a mistake here. The MIDI IN on the board itself that connects to the MPU-401 clone card, is the MPU-401 clone card connected from MIDI OUT to MIDI IN, then THRU to the UM-ONE? Is it the UM-ONE MKII?

bjwil1991 wrote:

For clarification, and please correct me in case I made a mistake here. The MIDI IN on the board itself that connects to the MPU-401 clone card, is the MPU-401 clone card connected from MIDI OUT to MIDI IN, then THRU to the UM-ONE? Is it the UM-ONE MKII?

That's correct.

The MPU-401 MIDI OUT is connected to the MIDI IN on the circuit. The UM-ONE MIDI IN is connected to the MIDI THRU on the circuit. The UM-ONE is connected to the RPI via USB. It performs exactly like if I would connect the UM-ONE directly to the MPU-401.
It is the MKII indeed. I have the MKI too.
This is only for testing the output from the circuit though. I intend to use the serial port on the RPI because this can be integrated on a PCB.

Cheers