Piping real-time audio between Asterisk channels and the outside (non-telephony) world is the core means by which we will hook up machine learning to the voice world. Explore the new mechanisms which make this fast and easy.
There are a number of situations in which performing real-time interactions with the audio streams of Asterisk is useful. Speech Synthesis, Speech Recognition, Tone interactions, fraud analysis, flow detection, and any number of machine-learning situations all require smart and realtime interactions with raw audio streams. In this talk, we will review the tools we have available for these interactions and demonstrate real-world implementations.