audiocraft-audio-generation

Load a pretrained MusicGen model with MusicGen.get_pretrained, set generation parameters like duration and temperature, then call model.generate with a list of text descriptions. Save the output tensor as a WAV file with torchaudio at 32kHz.

MusicGen generates music from text prompts and supports melody conditioning and stereo output at 32kHz. AudioGen generates sound effects and environmental audio from text at 16kHz. Both are part of Meta's AudioCraft library.

Yes, MusicGen is available in HuggingFace Transformers via MusicgenForConditionalGeneration and AutoProcessor. Load facebook/musicgen-small, process text inputs, and call generate with parameters like guidance_scale and max_new_tokens.

MusicGen-small requires about 4GB VRAM in FP32 or 2GB in FP16, medium needs around 8GB/4GB, and large needs roughly 16GB/8GB. Reduce duration or use the small model if you hit CUDA out-of-memory errors.

Silent output usually means the audio needs normalization or the generation parameters are off. Increase cfg_coef for better text adherence, lower temperature for more focused output, use a larger model, and write specific prompts describing genre, instruments, and mood.

Use Stable Audio for longer commercial music generation, and Bark for text-to-speech combined with music or sound effects. AudioCraft is best for controllable text-to-music, melody-conditioned generation, and neural audio compression with EnCodec.