Skip to content

Latest commit

 

History

History
51 lines (32 loc) · 1.38 KB

README.md

File metadata and controls

51 lines (32 loc) · 1.38 KB

Keyword Detection Demo

This demo is a server application consuming audio from Twilio Media Streams and using IBM Watson Cloud Text to Speech to perform realtime transcriptions and keyword detection.

App sever setup

Create a new IBM Cloud Resource

https://cloud.ibm.com/catalog/services/speech-to-text

On the Credentials card choose the Download link. Save the file ibm-credentials.env in the same folder as this file.

Installation

Requires Node >= v12

Run npm install

Running the server

Start with npm start any keywords you want to search for

Setup

You can setup your environment to run the demo by using the CLI (BETA) or the Console.

Configure using the CLI

  1. Find available phone number twilio api:core:available-phone-numbers:local:list --country-code="US" --voice-enabled --properties="phoneNumber"

  2. Purchase the phone number (where +123456789 is a number you found) twilio api:core:incoming-phone-numbers:create --phone-number="+123456789"

  3. Start ngrok ngrok http 8080

  4. Wire up an incoming call handler to your number using a TwiML Bin

<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Start>
    <Stream url="wss://<Your ngrok host name here>.ngrok.io/media" />
  </Start>
  <Pause length="40" />
</Response>

  1. Call your number +123456789 from a phone.

  2. Speak the keywords and see them get detected in realtime