Open Assistent only begins listening intermittently

[maxcr]

Half the time I begin Open Assistant the command prompt does not refresh. Instead I only get red < ? > I am assuming because the program defaulted to an incorrect microphone device.

When it does work it not recognize my speech very often. Receiving debug output from sphinx-pocket would be nice.

If you guys could get a Slack going I'd love to begin helping you with development. I'm running Arch on Kernel 4.8 with grsec and everything is working . Obviously this needs lots of development but I really want to help and there's no registration on your website.

[vavrek]

Greetings Max,
Thanks so much for trying out Open Assistant and for your interest in getting involved!
As you can see this is an extremely young project .
Will setup a Slack network as soon as possible and invite you in.
This is a great idea.
At the bottom of 'assist.sh' are instructions for dealing with microphone settings.
Perhaps this could help...
Will be in touch soon!
Andrew Vavrek
http://openassistant.org

[vavrek]

Also, we do have a forum just starting up now:
http://openassistant.org/community/

[vavrek]

Hey max...
Just setup a Slack network: https://openassistant.slack.com
If you send me your email address, I'll invite you in.
Cheers.

[maxcr]

[email protected]

Sorry about the slow reply been working on trying to get AMDGPU-PRO working for Arch Linux and... well you can see if you click on my github profile...

[vavrek]

Just sent you an invite!
Will be pulling in your install tips from your version of the README eventually...
Thanks for doing this.

[maxcr]

Any time brother

On Nov 5, 2016 9:26 AM, "vavrek" [email protected] wrote:

Just sent you an invite!
Will be pulling in your install tips from your version of the README
eventually...
Thanks for doing this.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#1 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABLRxgVRTk8tdbL3KtaFELcp894guIhPks5q7KAMgaJpZM4KYqNb
.

[Whytehorse]

This issue is caused by the dictionary(or lack of it). The speech recognizer sends text to a dictionary lookup. If nothing is found, it assumes it's not a command or garbage. There are currently only a few commands that are in there. When you load a full dictionary, you get wild results and slow performance. This bug could be solved by having OA display what it thinks you said and then either perform the command or display "unrecognized command". Something like that.