A speech recognition application leveraging Deep Java Library

by React

DJL demo

A small demo app that showcase building a speech recognition application leveraging Deep Java Library. The app is composed of:

  • Backend service: Spring boot app, loading wav2vec model and exposing a rest endpoint accepting an audio file and returning a string
  • Frontend app: Built using react JS, enabling to record an audio input and calling the rest endpoint. Both apps are for demo purposes only, and the code practices followed in this repo reflects that 🙂 PRs are welcome.

Running the app

Requirements

  • Java 17, maven, npm

Downloading deps:

First build your backend application using mvn clean package. For the frontend run npm i. Both assumes that you’re in the folder of each app.

Starting apps

run the backend using java -jar paht_to_your_jar, and run npm start to start the front in dev mod. head to http://localhost:3000 and have fun 🙂

Build using Docker

TODO: This can be simplified using docker compose.

GitHub

View Github