metadata

title: Fire Coml FALL 2022
emoji: 🎶
colorFrom: green
colorTo: indigo
sdk: streamlit
sdk_version: 1.10.0
app_file: app.py
pinned: false

Tag-based Audio Generation

Description

Model to generate audio given input genre(s), mood(s), and instrument(s)

Link to demo app

https://huggingface.co/spaces/SLAYEROFALL3050/Audio_Generator_Using_GAN

Youtube Video demo

TODO

System Architecture Diagram

Explanation

User inputs one genre tag, mood tag, and instrument tag into frontend. Each tag is passed to semantic similarity NLP model to determine nearest tags within training space, and coerces to (outputs) found training space genre, mood, and instrument tag. Those tags are passed to the audio generation model as input, which produces generated audio which is playable on the frontend.

Model Architecture Diagrams

TODO: NLP model diagram

TODO: Audio generation model diagram

Directory Guide

TODO

Training Instructions

TODO

Testing Instructions

TODO

Citations and References

TODO