@mswcpt I highly recommend trying a Rpi sound card. The ones I've tried sound so good. Better than my Motu interface on the Mac. I am using an Audio Injector one now. The PiDac+ sounds amazing, but it doesn't work well with other Pi shields, but sounds really amazing, Hi-fi. I recommend going that route. Set it up, in PD select is as your audio interface, save chagnes, and next time you launch Pd it should be the same setting.

@nicnut Thank you for the reply. We are going to try with an audio interface that works on Pi! It may very well be that systems that experience this error would benefit from external audio hardware controllers rather than on-board. Will post result once tested. Cheers

@mswcpt said:

@marivier.rocks Hi - did you get this reproduced on a new SD? I have this problem even after trying your solution. Thanks

My solution wasn't the good one, sorry. See hereafter others tries...

The "alsa xrun recovery apparently failed" bug seems to be solved here : https://github.com/pure-data/pure-data/issues/969

So, I downloaded the latest source code and installed Pd 0.51-4 on my Raspi (with Buster ->Linux version 5.4.72-v7+ (dom@buildbot) (gcc version 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04)) #1356 SMP Thu Oct 22 13:56:54 BST 2020)

I use the onboard soundcard (bcm2835), as far as there is no input, I uncheck "input device". When I tested sound, there was no "alsa xrun recovery apparently failed" error message.

But some minutes later in "Pd noGui or GUI mode" + "Raspi console or graphic mode", the error is still there.

Holla! I have the same problem, running Pd 0.51-4, Raspberry pi 3 A+ (Should be up to date, installed and updated yesterday), using it's internal sound card. Patch includes [comport] that fetches data from an Arduino. Still investigating, but it seem to work for a couple of hours, then I get that same message, "alsa xrun recovery apparently failed". Gonna try another sound card.

I previously had version 0.49.3 on Raspi 3B. After opening a patch and additionally opening the object overview window, it crashed. My guess is that the DSP opens 2x independent DAC output, which doesn't work due to the BCM2835's lack of hardware mixing. Current compiled version 0.53 (puredata.info) can represent this problem at least in the console, if you start it from there. XRuns occur, but the program does not crash. You clearly see, that ALSA restarts like crazy. You should also check in the second window whether the ALSA settings are correct and save them there if necessary. It's obviously a sound chip issue. For the object overview, however, I added a [loadbang]->[;pd dsp 0] message and saved it.
You can also turn off DSP before opening another patch (or subpatch!) that will automatically turn on DSP, which happens under the hood, e.g., the help patches. With these changes, so far, it works for me. Furthermore, a proper sound card with hardware mixing would solve the issue. You get the idea.

Hello,
I have a similar issue and I think I'm almost there to figure it out.
I've found this Miller Pucket Github's post : https://github.com/pure-data/pure-data/issues/969
But I don't know what to do with the code.

Can you help me please?

@Boris The fix mentioned in your link is 3 years old...... so if you have updated Pd since then that it is probably not the source of the problem as it will have been changed already in later releases.
The fix was made to pd/src/s_audio_alsa.c which is a file that needs to be compiled as the Pd application is built.
If you built your Pd from source then you could change the c file..... change the numbered lines in the link to match the code in the link, and then re-compile Pd.
David.