📤 Share
📝 Summary
Meta AI's unified codebase for generative audio research.
🏷 Tags
⭐ Rating
📖 Tutorials
📝 About This Tool
•AudioCraft is a single-stop code base from Meta AI for all generative audio needs, including music generation, sound effects creation, and audio compression. It simplifies generative audio model design by using a single autoregressive language model operating on discrete audio tokens from the EnCodec neural audio codec. The platform supports text-to-music and text-to-sound generation, enabling high-quality audio output from text prompts.
⚡ Key Features
•Unified codebase for music, sound effects, and audio compression
•Text-to-music generation with MusicGen
•Text-to-sound generation with AudioGen
•EnCodec neural audio codec for discrete token learning
•Single autoregressive LM for efficient audio modeling
•Open-source research framework
✨ Why Choose It
•Simplifies generative audio model design compared to prior work
•Elegant token interleaving captures long-term dependencies
•Single model handles multiple audio generation tasks
👥 Who Is It For
•AI researchers in audio generation
•Developers building audio applications
•Music and sound designers exploring AI tools
❓ FAQ
Q: What types of audio can AudioCraft generate?
A: It generates music, sound effects, and compressed audio from text prompts.
Q: Is AudioCraft open source?
A: Yes, the code is available on GitHub under Meta AI.
Q: What models are included in AudioCraft?
A: It includes MusicGen, AudioGen, and EnCodec.
