No luck, but that seems to be focussed on playback?
When running ./start-mycroft.sh audiotest everything works fine, and the plugin shows up as seen in the screenshot below. But as soon as it stops, it disappears, and running Mycroft doesn’t get it back. Does the test somehow use a different plugin? Or perhaps because of the virtual env mycroft is running in?
That is interesting that the audiotest works but Mycroft doesn’t.
I saw in your original logs that /etc/mycroft/mycroft.conf was not defined and hence skipped. Is that still the case? Can you post the output of cat /etc/mycroft/mycroft.conf
And have you made any modifications to ~/.mycroft/mycroft.conf? As this will override anything in /etc/mycroft/mycroft.conf
There was no /etc/mycroft/mycroft.conf but I created it myself (see this post). It now only contains the TTS tag.
Also, I did make changes to ~/.mycroft/mycroft.conf, but as they got overwritten I created the /etc one. It now only contains one tag, something like ‘max allowed version’ (not near the machine atm).
Is there supposed to be a conf in /etc by default?
I really dont know how this resolve it but it works for me. I use the option “all” locales and it take a bit long to finish the reconfiguration but mycroft started listening to me
About the configuration file, I found it in …/mycroft/configurations/mycroft.conf
Thanks @michaelmfzv for taking the time to respond I tried your advice; it’s true that I didn’t have a language set, so I ran the commands and tried both en-US and en-GB but nothing changed
And the following error message keeps popping up in audio.log, but I don’t think it’s related:
21:05:39.950 - mycroft.audio.audioservice:load_services:110 - ERROR - Failed to import module mplayer
ModuleNotFoundError(“No module named ‘py_mplayer’”,)
Is there no dev around that can at least point to a direction as to why recording works during the test, and not when running Mycroft? I’ve already logged considerable hours into this and I haven’t even written a single line of code
Sorry that it’s still not working, audio is a challenge with so many variables across machines and different configurations.
It would be good to try and specify the exact device that Mycroft should use through the config. We usually leave it set as essentially ‘system default’ but in this case that clearly isn’t working.
At the bottom of the output it should have something like:
==========================================
Found devices:
HD-Audio Generic: ALC892 Analog (hw:1,0): Samplerate 44100.0 Channels 2
HD-Audio Generic: ALC892 Alt Analog (hw:1,2): Samplerate 44100.0 Channels 2
pulse: Samplerate 44100.0 Channels 32
default: Samplerate 44100.0 Channels 32
Copy this somewhere so you can reference it.
2. Specifying the device in mycroft.conf
Exit venv if not already ~ $ mycroft-venv-deactivate
Edit configuration file ~ $ nano ~/.mycroft/mycroft.conf
Add the following lines:
"listener": {
"device_name": "DEVICE_NAME"
}
Where “DEVICE_NAME” can be pulse or in the example output above “HD-Audio Generic: ALC892 Analog (hw:1,0)”
3. Restart Mycroft
To make sure it’s fully restarted run:
$ cd ~/mycroft-core
$ ./stop-mycroft.sh
$ ./start-mycroft.sh debug
I’ve also flagged this post with one of our core dev’s to see if they have any other ideas about what might be happening particularly given discrepancy between the audiotest and real usage.
Since the release yesterday the audiotest has an option to list the available devices:
./start-mycroft.sh audiotest -l
which should be simpler than the script I published as a gist.
At this stage we might want to check the voice.log for any errors. The voice.log is found in the folder /var/log/mycroft/ Can you upload this file to pastebin.com or similar service?
Do you get the “Ding” indicating that the device has heard “Hey Mycroft”?
@gez-mycroft Awesome, that seems to have done it It defaulted to nothing, no mic or anything. Think that’s something that Mycroft can maybe implement in the installation or configuration script, to manually select a microphone as a default instead of using the system wide default.
Weird thing is though, that while I can see voice.log scanning for the mic I defined in the conf, it still shows this line at the end of it:
@forslund: I actually have more for you in case you get bored
I tried to switch from pocketsphinx to precise, but that didn’t go well as you can see here.
At 09:50:20.191 there’s actually an error, after that it seems busy recovering itself and now it’s constantly ‘still downloading executable’.
Mycroft indicates it’s using precise as the listener by the way, pavucontrol indicates Python is listening but there’s no voice response to ‘Hey Mycroft’. I’ll let it run for a while and after that revert to pocketsphinx.
Thanks for reporting. Seems like some error handling is appropriate here. I tried to reproduce but couldn’t. Maybe a broken download but it’s a bit weird since the downloader should verify the checksum before extracting… I’m going to flag @Wolfgange to see if he has any suggestions.
In the meantime I suggest removing the ~/.mycroft/precise folder and restart the voice service to see if the a redownload solves the issue.
Went back to pocketsphinx, shutdown Mycroft, deleted the ~/.mycroft/precise folder, restarted Mycroft and attempted to switch to precise - but same thing happened and it’s again in the ‘still downloading executable...’ loop.
An update on the previous issue with it not selecting the correct device. The device_name is a used as a regex match so the parenthesises are messing it up. It should work if you just enter HDA Intel MID: ALC270 Analog into your config.
@forslund: jep that fixed it, it’s picking the right mic now
@j1nx: tried that but doesn’t appear to be working, however when asked it says it’s using precise
Another new bug is that some output gets printed in the terminal instead of the tool, I think it’s when remote config gets fetched… After a few sec the tools removes it so no biggie:
Unfortunately, this still isn’t tackled. After a few busy days at work I booted up the Mycroft machine, only to find that recording once again isn’t working - nothing showing up in pavucontrol. Tried rebooting, checked to see whether the settings were still in mycroft.conf - they’re there.
I pointed the listener to precise as you can read above here, but that never worked. I’ve set it back to pocketsphinx but no difference.
Weird thing in the voice.log is, it just stops in the middle as you can see here. Seems to be halting on something?
Do any of you have any idea of what’s so different about this system - it’s Ubuntu Mate 18.04 with a Blue Snowball mic. Vanilla install and dedicated laptop, nothing else has been done with it.