Pyaudio random failures with 2.5.0-pre

Hi all,

I’m using 2.5.0-pre in Docker and experienced the following errors after several hours running Rhasspy (even in the middle of the night without any ongoing activity). Note that it happened on Pi-zero and several hours later on Pi-3B+ as well.

Both PIs are connected to a Jabra 410 but not sure that it has any link…

    [ERROR:2020-05-27 03:22:09,556] rhasspymicrophone_pyaudio_hermes: record
Traceback (most recent call last):
  File "/usr/lib/rhasspy-voltron/rhasspy-microphone-pyaudio-hermes/rhasspymicrophone_pyaudio_hermes/", line 121, in record
    chunk =
  File "/usr/lib/rhasspy-voltron/.venv/lib/python3.7/site-packages/", line 608, in read
    return pa.read_stream(self._stream, num_frames, exception_on_overflow)
OSError: [Errno -9981] Input overflowed

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/lib/rhasspy-voltron/rhasspy-microphone-pyaudio-hermes/rhasspymicrophone_pyaudio_hermes/", line 128, in record
  File "/usr/lib/rhasspy-voltron/.venv/lib/python3.7/site-packages/", line 524, in stop_stream
OSError: Stream not open

Is it known or do I have to open a github issue?

i might have been experienced the same problem ( but i reinstalled everything and now since the wake word doesnt work cant say )

i used to reboot my rhasspy every 5 hours so it doesn t stay offline very long

I don’t have this in my logs with the ReSpeaker 2-Mics Pi HAT. Seems to be a problem specific to the Jabra device then? Are you using the same device, @lilbuh?

no i have a pseye 3 hooked up on my sattelite
( sorry i have to make a post of 20 words minimum soignore this line :stuck_out_tongue: )

Is it something to do with powersave? Being USB?
Dunno but just a suggestion that its dying after a period of non use?
Not that I am really sure but it could be a maybe.

Yes that could be it, both devices being USB based.

Do you see any strange messages (especially error messages shown in red) in the output of dmesg, @tuxedo78 and @lilbuh?

the only weird messages i can see right now ( and i did have them before wiping everything ) is those

[ 1675.165354] usb 1-1.3: current rate 4533353 is different from the runtime rat e 16000
[ 1686.724896] usb 1-1.3: current rate 92449 is different from the runtime rate 16000
[ 1844.326856] usb 1-1.3: current rate 1260788 is different from the runtime rat e 16000
[ 1870.093281] usb 1-1.3: current rate 206104 is different from the runtime rate 16000
[ 1886.202987] usb 1-1.3: current rate 128841 is different from the runtime rate 16000

and the raspberry pi on raspian dont have powersaving features at all

PS3eye does that all the time, dunno why its gets recommended so much as the alsa controls are not right either.
It works OK.

Dunno think it might be the desktop that has powersave options. So dunno.

That seems to be a common message with USB sound devices:

PS3eye crops up a couple of times in that 1st page and its cruel for it.

I have been running a few USB sounds cards due to EC testing and maybe because I am trying to get ones with a stereo ADC they are a little better than average but the edimax without postage was only £5.
Never get any such messages but do at times get a strange one about it waking up with no data but nothing happens and haven’t a clue why pulse recommend contacting alsa.

Think its cheap old webcams and soundcards but most don’t do such things.

Hi @koan, I have rhe same issue as @lilbuh has
and it happend after full reinstall 2 weeks ago or so
in dmesg i see some red lines

[   22.201745] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac43455-sdio for chip BCM4345/6
[   22.202609] usbcore: registered new interface driver brcmfmac
[   22.204200] gspca_main: ov534-2.14.0 probing 1415:2000
[   22.463423] brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac43455-sdio for chip BCM4345/6
[   22.508741] brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4345/6 wl0: Mar  2 2020 23:30:41 version 7.45.202 (r724630 CY) FWID 01-72f6ece2
[   25.200505] usbcore: registered new interface driver ov534
[   25.322222] usb 1-1.2: current rate 184185 is different from the runtime rate 16000
[   25.324942] usb 1-1.2: 3:1: cannot get min/max values for control 2 (id 3)

but according to my tests

  • Snowboy is not working… just doesn’t wake up
  • Porcupine works, but I do not hear wake and recorded sound
  • Wake button works, but with no sounds too
  • Play button works(!) and I hear my voice well

so i guess it’s not hardware issue. Rhasspy recognizes my voice, and can play it

maybe a stupid question, but is it possible to download NOT the latest docker image? maybe april one…

That is another PS3eye isn’t it as there does seem to be a pattern here and on the web with the ov534 driver.

I’m not sure you can, because they are all published with the same tag 2.5.0-pre.

Yes, it is. I know you hate PS3eye, but…
-it’s cheap
-it looks good staying on my desk
-it’s fully supported by RPi, just plug and play
-It has embedded camera and I use it
-It’s more than enough for me, and have been working since Snips

ohhh, ok
will investigate further

Goggle how many errors it causes it has a titanic history of problems.
Its a piece of hardware that was designed for a console that only works on a PC because the silicon available was PC stock.
Its not fully supported its driver is an incomplete reverse engineered hack that can be proven by doing a simple alsactl store.

If your ever ill, please for your own good don’t use similar choice methods for remedies, for you own benefit.
Use your preference it is no bother but don’t expect any support as its got nothing to do with hate as that would be purile and illogical.
It causes problems and it wastes time as it has in many projects where its been used as its history is documented through what can be returned in a Google.

Basically its c*** and you wasting energy by even posting about it.

I have one on my desk that remains from my initial exploration of Mycroft and thier recommendation and the only reason its not in the bin is because I am a terrible hoarder and hate waste.
I do hate waste and really it was buying it.

It was exceptionally good on the PS3 for time and cost its reuse on the PC is not so and it seems an urban myth based on previous glory and that it has 4 mics that create far more problems than benefit.

The kernel module snd-usb-audio is not great at the best of times but in conjuntion with that device its actually quite awfull to control, there are a few webcams of that time that are still a bit flakey and really sensibilly should be avoided.

@rolyan_trauts Almost every your message is about how PS3eye is bad. Ok I got it :nerd_face:
But you should admit that different people have different needs, and basic home usage is not the same as sound projects you failed because of PS3eye

So if you want to help solving the issue, you can completely remove 2.5pre docker container and image, then get fresh one, and check whether it works for you
If you don’t… just ignore my messages. It’s easy.

There is no solving the PS3eye and the project should definately not revert back on the premise of one users wish to use a historically broken device.

You can ask like you can use the PS3eye and see where it gets you.
If there is an update that helps it limp along then hey go for it, but users should be aware its always likely to cause problems that are not inherrent with other devices in use.

Stating that all problems are caused by PS3eye won’t help the project. Especially when you have no evidence except “go google it”.
Just try to live with the fact that many people still use it, and don’t be so emotional.

There is no emotion or niether I am saying all problems are caused by the PS3eye.
Just that PS3eye problems are a waste of the projects time.

So may we “PS3eye users” post some our problems here? ok?
If you see real hardware issue, for sure, your help will be appreciated
But if you don’t, I think there is enough such comments across the site: