Having trouble registering/pairing my microphone

I’m on Linux Mint 18 with a working samson mic, got everything installed via the github page, rebooted, issued ./mycroft.sh start – service, skills, voice, cli appeared to start normally, entered the registration code at Cerberus, I have a “prototype” paired device listed and a normally-looking identity.json local file with the device ID matching. But, no matter how many times I’ve restarted mycroft, the voice still complains my device isn’t registered yet and gives me a new code to register – which upon registering again still leads to the same device.

There’s 1 error visible screen -r mycroft-service here’s the output upon starting:
requests.packages.urllib3.connectionpool - INFO - Starting new HTTPS connection (1): cerberus.mycroft.ai
mycroft.configuration - ERROR - Failed to fetch remote configuration: ValueError(‘No JSON object could be decoded’,)

Connection error? The same url appears to be to working in the voice screen output…except perhaps this line:
requests.packages.urllib3.connectionpool - DEBUG - “POST /stt/google_v2?language=en-US&version=dev-8846268 HTTP/1.1” 401 75
mycroft.client.speech.listener - ERROR - AccessDenied from Cerberus proxy.


I seemed to have solved this issue mysteriously. I de-registered the existing device and saw there was a couple more scripts to allow in my browser extension, NoScript – so I enabled them all, and then proceeded to re-register the device and now it appears to be working. And apparently I seem to be in Kansas :neutral_face:

Quick question: is there a quick voice command to tell Mycroft to “shut up” in the middle of a very long rambling definition of the word “Oregon” it thinks you said after deciphering “what’re ye’ doing?”

“Hey Mycroft, shut up” would be nice.

I’m just going to continue with this thread for my own feedback. One annoying thing, along with the ability to not being able to stop Mycroft from reading a long dictionary reference to “Oregon,” is the ability to configure it to not dictionary word-search the first word it doesn’t understand in a string.

For example, the voice input, “Hey Mycroft, do you conspire with the NSA?” – results in a long definition of the word “conspire.”

For telling him to shut up, Mycroft stop should work for some skills. It’s not entirely configured for a lot of them, though.

Part of the problem with the long definitions is that if Mycroft can’t match what you say with one of the skills, then he’ll fall back on Wolfram Alpha. So apparently Wolfram interprets do you conspire with the NSA as conspire. I have a PR open that will fix some of the long definitions by limiting it to speak just one definition rather than something like 11.

What’s the point of even having a definition read back if the intention of a definition wasn’t explicitly there? For example, if the user input is misunderstood, why is that be interpreted as being, “define the evaluation I misunderstood, with the assumption that the user requested a dictionary definition.” See where I’m going with this?

At the moment, Wolfram Alpha is treated as the fallback so that it can more easily handle a lot of queries, and so that we don’t have to try to specify that it can handle something like what is 2 to the 3rd as well as something like how far is it from the earth to the moon. Unfortunately, this also results in a lot of bad results, as there are a lot of misinterpreted queries that go through to it as well.

Also, the assumption is not that the user requested a dictionary definition. The assumption is that we don’t know how to handle it with our current set of skills, so we send it to Wolfram, who parses it as wanting a dictionary definition.