How to Set Up Wildix Speech to Text

This document explains how to set up Wildix Speech to Text functionality

Created: October 2022

Permalink: https://wildix.atlassian.net/wiki/x/AYDsAw

Introduction

Speech to Text (STT) functionality allows to create an automatic system that asks a caller some questions, recognizes and transcripts the answers and lets you save the recorded files. This could be quite useful, if, say, you receive lots of incoming calls, and to handle them all, you set up such a self-service system. Callers leave their data via IVR, the data is then saved to JSON and passed to the database. 

Using a script 

To have all recorded data saved into one file and passed to a database, besides configuring Speech to Text application in Dialplan, you need to use a script. The script:

  1. Receives the JSON file from Dialplan variable RECOGNITION_RESULTS
  2. Inserts the data from JSON to customer’s database
  3. Downloads audio files if available, concatenates them into a single file with 2-second pause between the recordings
  4. Removes all unnecessary data (e.g. audio files) from the PBX.  

Setting up STT

  1. Download the Wildix STT script and see examples of its usage by the following link: https://github.com/Wildix/example-script-STT 
  2. Configure Dialplan application Speech to Text. Documentation: Dialplan applications - Admin Guide: Speech to Text