Picroft aborts speeking since 19.02 update

No change on the aiy-cutoff attempt.

Thanks wiggins. Back to the drawing board I guess.

Running 19.2.12 freshly updated from the stable image and still seeing the same thing (AIY hat build).

@forslund: I made a interesting observation today on my Picroft with AIY voice kit:

When - again - Mycroft aborted speaking I pressed the LED-button on top and suddenly Mycroft continued to speak the whole text (for several minutes as it was a very long text).

1 Like

I’ve been trying to look at the aiy-skill to see what that could mean but I see no real reason here. A short press should send the mycroft.stop signal, which should abort the speech by killing the aplay process… So this is definitely weird.

Do the setup for tha aiy voice kit use aplay or paplay for audio playback?

playback: it is aplay (in /etc/mycroft/mycroft.conf)

I am just wild guessing here: maybe audio output of the AIY-voice-hat gets muted for some reason and pressing the button un-mutes it again?

You’re probably right about the mute but what is causing it?

Btw @steve.penrod pushed a new unstable image this morning, it might be worth a shot to see how well that image is working.

Talked with Steve today and he actually added a google aiy specific fix for chopped audio: https://github.com/MycroftAI/enclosure-picroft/blob/stretch/home/pi/auto_run.sh#L328

Would be great to have some feedback on this.

1 Like

commented out load-module module-suspend-on-idle in `/etc/pulse/default.pa´

…Woo-hoo - wikipedia star wars followed by continue - Picroft talks and talks and talks…

Thanks.

2 Likes

Confirmed fixed here, too. Just to be clear, you have to make the change mentioned by @Dominik.

When I followed the link from @forslund and then the link on that page (See:) The last line in the thread says

This should be fixed on the latest SD card image.

I thought that meant it would be fixed in the image on the Picroft GitHub page (Picroft Stable 2019-07-10 image) but was wrong.

1 Like

Yes! This worked for me! I had to upvote in some way :slight_smile:

1 Like

N00b here, downloaded and used the .zipped image hosted by Mycroft on Google Drive, linked from that Github page, and am experiencing this same problem.

Is it Picroft Stable 2019-11-10 image you downloaded?
The old coments mentions older version of the Picroft image, and as far as I can tell it is fixed in the latest. 2019-11-10

Yes, I downloaded and installed it last week.

Hmmm I did to and it is working for me.

But I always firat exiting the auto_run script and perform

sudo apt-get update
sudo apt-get upgrade

to make sure everything is updated. And then reboot and walk throu the setup.

Is it only sound that are missing ? Can you see mycrofts speach in the mycroft-cli-client ?

Yes, I can see Mycroft’s responses to voice inputs perfectly well in the cli client. In fact, if I leave it on for a day or more, I’ve now found that the problem worsens to the point where no voice/sound at all comes out of the speaker. I’m not married to Google hardware, I’ve only been using the kit because I found it at a discount – so I’ll likely swap out the AIY hat and output device for something else. Will hunt at Goodwill store for cheap speakers. I’m not clear on something, though, after reading the documentation: If I come across a USB speaker, how necessary is it to also use a USB sound card of some kind?

OK then mycroft is working, but hardware/setup is not.

if you use alsamixer can you encrese volume and mute/unmute ? And what outputdevice does alsamixer say you use ? (should be pulseaudio)

I’m afraid I never checked, I just did the setup wizard, selecting Google AIY for both input and output, and left everything else (vis a vis outputdevice) as whatever the default is. For what it’s worth, whatever I was using was able to adjust the volume up/down.

I tried installing the Linux version from the master branch and it the problem still exists.

Right, so I got the same problem on my picroft (Pi 4) with an AUX Speaker.
I’ve tried pretty much everything mentioned in this thread - still no change. I can speak commands that get answered - in the cli, but not via audio. The listening-sound plays though. What else can I do?