Via The Neuron
Meta Open-Sources Audiocraft, It’s Text-To-Audio Tool 
|
|
|
| “AudioCraft is like ChatGPT but for audio: generate any type of sound imaginable. |
| Meta just open-sourced the entire damn thing (read: it’s now available for me and you). Here is what’s possible |
- create any type of music, melody, or soundtrack.
- create any sound effect or noise.
- break down any sound into its smaller pieces.
|
| Hands down the best AI model for audio (sorry, Goog). |
| So what’s the big deal? |
| First, music changes in a big way. Anyone can customize a song, composition, or melody simply by describing it. Try this formula: “a [type of music] track with [type of instrument] at [#] BPM in the style of [musician’s] [specific melody/song].” |
| Second, it becomes wayyyy easier to create sound effects for things like: |
- films.
- games.
- podcasts.
- video content.
- ad jingles. I’m lovin’ it!”
|
| Why it matters: Meta delivered, and they did it with grace: every second of the 20,000 hours of music used to train AudioCraft has been licensed. Spect. |
| We’re about to see some truly innovative projects and products built on top of AudioCraft. Here’s the code if you’re a builder. Share your creations with us! |
0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.