search

alexa-pi / AlexaPiDEPRECATED

DEPRECATED - Use https://github.com/alexa-pi/AlexaPi instead ...Turn a Raspberry Pi into an Alexa Client
https://github.com/alexa-pi/AlexaPi
MIT License
587 stars 439 forks source link

Without button to listen? #50

Closed sabas1080 closed 8 years ago

sabas1080 commented 8 years ago

Hi

Is there any option to not press button and Alexa listen?

Use voice command "Alexa ..."

Thanks

GemBro commented 8 years ago

At the moment no ... not without breaching the Amazon terms & conditions ...

robertlacok commented 8 years ago

Can you please link the relevant T&C? I read through the ones I could find and never saw one about having to use a button. I would like to use jasper to listen for the trigger word and record the command and then pass it on to AVS.

Edit: found them, https://developer.amazon.com/edw/avs_requirements.html they show once you are registered as a developer.

doubledrat commented 8 years ago

I'm sure I've seen it somewhere and someone else said so too. I can't seem to find anything now though. Perhaps they've removed that requirement?

On Fri, Apr 15, 2016 at 8:07 PM Robert Lacok notifications@github.com wrote:

Can you please link the relevant T&C? I read through the ones I could find and never saw one about having to use a button. I would like to use jasper to listen for the trigger word and record the command and then pass it on to AVS.

— You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub https://github.com/sammachin/AlexaPi/issues/50#issuecomment-210600702

cheers Bob

tube0013 commented 8 years ago

There is a project that looks like it forked Sam's code, and has voice activation: https://www.hackster.io/fcooper27/beagleman-56f174 https://github.com/fcooper/beagleman

I tried integrating the code for sphinx into the lenny-music-1 patch but could not get it to work.

sammachin commented 8 years ago

I've done some testing with wake word detection, putting aside the T&C's issue for now;

Jasper seems to use google voice for speech rec whcih means everything that is said in the vicinity of your Pi is being streamed up to their cloud, this doesn't work for me from a bandwidth or privacy perspective.

I did look at Pocket Sphinx which is based on the same CMU sphinx project that beagleman uses however I found it had avery high false positive rate which also made it impractical for real world use. I'd like to try Sphinx a bit more, if anyone fancies creating a simple project (ideally in pyhton) that just listens and lights the LED for 5 secs when it hears the wake word that could be good for testing.

Oh the other problem is that you also have to do silence detection to figure out when the user has finished their command but this is easier to implement just might need tuning for certain scenarios