LeeBoonstra.dev

A best practice for streaming WebRTC audio from a browser microphone to Speech APIs

Type: Keynote / Breakout Talk
Time: 25min - 50min
Level: Advanced
Audience: Conversational AI practitioners, Developers

The Google Assistant is super popular. Businesses want their own voice AI. Unless you are a manufacturer for tv setup boxes or headphones, you probably don’t need to use the Google Assistant, you can integrate a (voice supported) conversational AI within your own existing web/mobile app with the use of Dialogflow and Cloud Speech to Text (STT), Cloud Text to Speech (TTS).

However, implementing audio streaming within (web) applications can become difficult. It’s not that the Google APIs are hard to use; those APIs are pretty straightforward. The problem is the combination of all required technology; A front app, WebRTC, microphone streams, audio encoding, sample rates, websockets, a back-end app, Google streaming APIs…

Lee Boonstra, developer advocate for Conversational AI, (and author of the ultimate audio streaming to Speech APIs - guide), will guide you how to implement your own conversational AI within your web application.

Message for the committee:
Expect a session, with a real world demo (Airport Self Service Kiosk), code to stream microphone audio output (WebRTC) to a back-end web application over websockets, and let your conversational AI speak out the answers. A solution built with: Dialogflow for chatbots, Speech to Text, Text to Speech, Websockets, Angular, Node.js, Audio Streaming, WebRTC

Disclaimer: The opinions stated here are my own, not those of my company. - © 2020 Lee Boonstra