Web Client Quick Start

The official Speechly client libraries for various platforms enable you to get up and running with Speechly quickly.

Web Client Quick Start

  1. In order to install Speechly browser client you’ll need to install some common developer tools. These include yarn and NodeJS . You’ll also need a Speechly app ID that you get by signing up to Speechly Dashboard.

  2. Clone the speechly-browser-client Github repository to your home directory and move on to the directory where the browser-client is cloned.

$ cd ~ && git clone https://github.com/speechly/browser-client/ && cd browser-client
  1. Install depencies required for the Speechly Browser client by using Yarn.
$ yarn install
  1. Move on to the examples directory that contains the web app and install dependencies for the examples, too.
cd examples && yarn install
  1. Get your application ID from the Speechly Dashboard. Make sure your application status is Published.
Speechly Dashboard listing Your Applications
Application ID is circled in blue. Note the Status column, too.
  1. Add your Application ID and model language to environmental variables. Your app language is probably `en_US, but you can configure this, too in Speechly Dashboard.
$ export REACT_APP_APP_ID="your-app-id" 
$ export REACT_APP_LANGUAGE="your-app-language" 
  1. Start the web application.
$ yarn start
Speechly Web Client example app starting in terminal
You should see Speechly web client starting on your local machine.
  1. If your browser doesn’t do it automatically, open your browser and navigate to the address visible in the terminal window. This address is probably http://localhost:3000

  2. After allowing your browser to use microphone, you can click and hold Record button and say utterances that your model understands. Once you start speaking, you should see the tentative transcript, tentative intent and tentative entities until those are finalized as per Speechly SLU loop.


Profile image for ottomatias

Last updated by ottomatias on March 2, 2020 at 14:07 +0200

Found an error on our documentation? Please file an issue or make a pull request