audiocraft-audio-generation

To generate music from text prompts using AudioCraft, install the audiocraft package, load a pretrained MusicGen model, provide your text description, and save the resulting waveform as a WAV file for immediate use.

Yes, you can create sound effects from text descriptions with AudioGen. It generates environment and sound-effect clips directly from natural-language prompts, allowing you to produce custom audio cues without searching for samples.

AudioCraft requires installing compatible versions of PyTorch and Transformers, specifically torch>=2.0.0 and transformers>=4.30.0, along with torchaudio, to ensure the neural codec and text-to-audio generation models function correctly.

The best way to encode and decode audio using a neural codec is utilizing AudioCraft's EnCodec workflows. EnCodec provides high-fidelity neural compression for encoding and decoding audio, enabling controllable processing pipelines.

Yes, MusicGen supports melody conditioning and stereo output. You can generate short to mid-length tracks from text descriptions, including melody-conditioned variants and stereo models for richer audio generation.