🎶 What is MusicGen?

Incredibly entertaining and thought-provoking.

Hello Everyone,

In a nutshell, I think MusicGen is pretty cool.

Will A.I. augment creativity?

Felix Kreuk, an AI research engineer at Meta, showed off the “MusicGen’s” capabilities in a Twitter thread late last week.

Meta A.I. is doing some really neat work on open-source A.I.  All credit to Jade Copet, Felix Kreuk, Tal Remez, Itai Gat, David Kant, Gabriel Synnaeve and Yossi Adi.

The sample posted here on LinkedIn recently, blew me away.

It made me laugh hearing the Palantir CEO talk about LLMs as poetry-generators.

The LLM-poetry quote happens at: 2:45 of this video. I don’t know why, but it makes me laugh.

Palantir AIP looks pretty dangerous frankly. But seriously, AI has managed to intrude on most artistic endeavors, and now it’s fully come for the music industry perhaps.

Speaking about poetry, how about music eh?

Text to Music is here in 2023

• Listen to our samples: https://ai.honu.io/papers/musicgen/

• Read the paper: https://arxiv.org/abs/2306.05284

• Play with our demo on HuggingFace: https://huggingface.co/spaces/facebook/MusicGen

MusicGen

Meta's Audiocraft research team has just released MusicGen, an open source deep learning language model that can generate new music based on text prompts and even be aligned to an existing song.

See the Tweet of Felix here.

MusicGen is basically a simple and controllable music generation model. MusicGen can be prompted by both text and melody. They release code (MIT) and models (CC-BY NC) for open research, reproducibility, and for the music community: