AI GitHub
AudioCraft for music and sound generation research
Experiment with published generative audio models and training components.
Key Takeaways
- AudioCraft includes model implementations such as MusicGen and AudioGen, along with code for running selected workflows.
- Choose the model for music or general audio and obtain its required weights.
- Generate a short sample with a clearly defined rhythm or sound event.
- Model weights may use terms distinct from the code.
How it works
AudioCraft includes model implementations such as MusicGen and AudioGen, along with code for running selected workflows. The model choice determines whether prompts target music or general sounds.
What you need
Choose the model for music or general audio and obtain its required weights. Save generation settings and keep an unprocessed copy before normalizing or editing the result.
What to try first
Generate a short sample with a clearly defined rhythm or sound event. Compare prompts using the same model and listen through the full clip.
Evaluate one prompt with sparse sound and one with overlapping events. It is easier to spot prompt-following errors when the expected events are listed before listening.
Where to look for failures
Model weights may use terms distinct from the code. Check artifact-specific terms and avoid assuming a short result will remain coherent over longer durations.
Related Franklin pages
Keep generation and editing as separate reproducible steps. A waveform or processed clip should have a reference to the raw model output and its settings.
- Apply audio effects and load supported plugins when editing model-generated audio. Pedalboard for processing generated audio in Python
- Generate speech and experiment with reference audio and documented expression controls. Chatterbox for controllable text-to-speech
For an application where the quality of recording and handling audio matters, see Franklin’s Open Mic AI profile. Open Mic AI
Project documentation
Try it for your workflow
Explore the project’s documentation for setup and usage, or save this page to a collection for later.
Read the project documentation ↗