Real-time transcription and sentiment analysis of audio streams; on the phone and in the browser (Inglés)


Machine Learning


Salón 203
Domingo 10, 10:00
Fecha y hora

Discover how you can use Artifical Intelligence to perform sentiment analysis of an audio stream, in real-time! In this workshop, we'll show you how to use AI and NLP to figure out what a person is calling about and how they are feeling, all from a telephone call audio feed via WebSockets.


"In this workshop, we're going to learn how to create a virtual rapporteur. A digital assistant who can join any conference call; record it and provide participants with real-time insights into the overall tone of the call. Once the call is complete, we'll look at how we can use the call recording to provide participants with a text transcript as well as meta information about the call such as the most talked about concepts, keywords and entities.

Workshop outline

Creating our conference

  • What is Hug
  • Creating a conference line with Hug & Nexmo
  • Running our Hug server

Adding our bot and recording the call

  • Creating our bot API endpoint

Sending our audio to IBM Watson

  • Introducing Tornado
  • Handling a WebSocket Connection
  • Using on_message to proxy data to IBM Watson

Performing our sentiment analysis

  • Sending our transcription to the tone analyser service

Displaying in the browser

  • Demo

Fetching our recording

  • Downloading the mp3 with requests
  • Sending the mp3 to be transcribed

Using the Natural Language Understanding API

  • Sending our transcription to the NLU API
  • Saving the response

Pulling it all together

  • Final Demo"