Amazon instructions vs. AlexaPi Implementation

Hi all! (Heads up – not an issue, just a question)

I was trying to create my own version of an Alexa interface on RPi through Python (for learning!) using this implementation as an example. I noticed that on Amazon’s instructions regarding interfacing with Alexa, they say to open and maintain an HTTP/2 connection that remains open through the lifetime of the connection and is used to send/deliver events/directives.

However, in this AlexaPi implementation, it seems as though that step is omitted, and every time an command is given, the audio file is directly sent to the speech recognizer 'https://access-alexa-na.amazon.com/v1/avs/speechrecognizer/recognize'.

Could someone explain why the constantly open connection is skipped? Thanks!

[gregwa1953]

@David Cai, “Works in progress” are rarely public. That’s because some people would try to build it without it being finished and be very frustrated.

On Thu, Jun 23, 2016 at 8:17 PM, David Cai [e-mail] wrote:

Is that work in progress public? Would you mind pointing me towards it so I can have a general guide to look at? Thanks!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/sammachin/AlexaPi/issues/100#issuecomment-228239960, or mute the thread https://github.com/notifications/unsubscribe/ARRIMQE_RzDV9pwB7ht3bCZDucHCW0wIks5qOz4rgaJpZM4I9P8f .

[flooie]

@renekliment

I built an objective-c version as an extension of the work here that takes both voice commands and text inputs (so I can give it commands coffee shops). I wasn’t sure if was appropriate to let y’all know about it here in the python project. It works pretty well just need to add your own client_id - token - and secret. Would love help improving it.

https://github.com/flooie/Mecho