I have an embedded system with the purpose of recognizing physical gestures. What I would like to do is send Amazon servers a pre-generated voice input whenever a specific gesture is recognized. I'll have to interface between C and Java in some way to combine the two gestures with the Alexa java app, but I need to first figure out if it is possible to substitute the voice input from a different source.
I can envision a (albeit dirty) solution where the raspberry pi switches to a different input audio channel than the mic, the pre-generated audio is "played" while the app is listening, and then the audio channel is returned to its normal state, so that basic functionality is maintained. I would love to hear any other ideas!
I am mainly looking to find out if these ideas are in any way viable with Alexa. This is in some ways just a workaround to access the natural language processing of Alexa for use with gesture recognition. I am currently planning to work with the sample app for RPi, but I not certain that I will be sticking with method of accessing Alexa.
I appreciate any insight provided to help me decide how to implement this. Thanks.