Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Dialogflow QUERY TIME EXCEEDED #5

Open
VladoPortos opened this issue Jun 12, 2019 · 7 comments
Open

Dialogflow QUERY TIME EXCEEDED #5

VladoPortos opened this issue Jun 12, 2019 · 7 comments

Comments

@VladoPortos
Copy link

Hello,
I have managed to setup everything how ever Dialogflow is giving me trouble. I call to Peeqo and it wakes but anything I say returns red ring and in debug console it say:
UNPIPING DIALOGFLOW > QUERY TIME EXCEEDED

Is there some text log that peeqo does where I could look in for more details ?

@VladoPortos
Copy link
Author

Issue seems to be in dialogflow.js here:

	this.stream.once('pipe', () => {
		console.log('PIPING > DIALOGFLOW')

		self.unpipeTimer = setTimeout(()=>{
			console.log('UNPIPING DIALOGFLOW > QUERY TIME EXCEEDED')
			self.sttStream.unpipe(self.stream)
			mic.getMic().unpipe(self.sttStream)
			self.unpipeTimer = null
		}, self.listenFor)
	})

Unfortunately I can't figure this out yet, node.js is very unfamiliar to me so I'm not sure what is this.stream.once, cant find equivalent to it on node.js example for dialogflow api so I'm at loss.

The behaviour is, that after Peeqo responds to key word, and listen to my command... it immediately ends at this point. Maybe the API returns some error but its not displayed ?

@VladoPortos
Copy link
Author

I moved with the issue a little :) the debuging from console needs to be changed since when you try to kill the peeqo process with kill -9 the lxde will fire it right back up and when you start it manually with debug console than its showing different errors like above. Now the peeqo is working, but most of the phrases are not recognized by Dialogflow, maybe because I'm not native or something... would be nice to be able to train in to my voice....

@VladoPortos
Copy link
Author

Well how many ways is there to sa Hi ? Either dialogflow is extremely bad at recognizing words ( I have no problem what so ever with google assistant ) or something else is wrong :-/ ( I kind of suspect that it is not listening long enough or start listening too soon or something like that )

@VladoPortos
Copy link
Author

Ach this is so frustrating, it get like one word from 50 ... :-/

@shekit
Copy link
Owner

shekit commented Jul 1, 2019

this is almost always a connectivity issue. I faced the same issue and when connecting to a more robust network the issue went away. We are looking at implementing an offline speech to text engine by snips.ai . That way we wont need to rely on dialogflow. The initial results are promising but we are facing a few issues with the audio output. Snips.ai takes over the audio output and prevents audio from playing from within the app. If we can figure that out, we'll make the switch

@ascg
Copy link

ascg commented Jul 13, 2019

For anyone having the same issue: I made two changes that helped mitigating this issue a lot (there are still situations where I get the QUERY TIME EXCEEDED but it got far better):

  1. In electron/app/js/intent-engines/dialogflow.js increase the this.listenFor value (should be line 58 in the current version). Something around 5000 did the trick for me.

  2. If you are not living in the US, add country=DE (where DE should be replaced with your country code) in /etc/wpa_supplicant/wpa_supplicant.conf to increase Wifi stability.

@VladoPortos
Copy link
Author

Thanks guys, I'm in the middle of renovating so Peeqo is sitting in the box right now, I noticed that other project I had running was causing drops on network so that might have been an issue, I will resume as soon I have somewhere to live again :D

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants