Generative AI Audio with Audiocraft 3 – Audiogen with Gradio

About

This tutorial series is on making generative music and audio from a simple text description of the sound using Facebook's Audiocraft Library.

In the last video, we set up a web form to be able to submit descriptions to a python backend, and return the generated audio file for download.

In this video we'll replace our complicated front and backend with some very simple UI components and functionality using Gradio, the interface commonly seen used in Hugging Face (https://huggingface.co/spaces/facebook/MusicGen)

Starting code: https://github.com/heaversm/audiocraft-tutorials/tree/tutorials/03-audiogen-flask

Final code: https://github.com/heaversm/audiocraft-tutorials/tree/tutorials/04-audiogen-gradio

About Me: I am a Staff Design Technologist on Mozilla's Innovation Team. All opinions and explorations are my own. Learn more about Mozilla Innovation at future.mozilla.org

0:00 - review 0:45 - gradio interfaces on hugging face 1:30 - remove all the complicated frontend and backend code 2:00 - installing gradio 3:25 - creating a simple gradio frontend 4:45 - understanding gradio blocks and components 5:40 - running the gradio app 6:20 - setting up our form, submission, and audio player with gradio 7:30 - modifying our generate audio function to work with gradio 8:35 - testing our final app

9:20 - wrapping up and next steps

Human-made No record of this work says a model made it. What this means.