audiocraft-audio-generation

Generate music from text prompts using PyTorch by configuring Meta AudioCraft transformer components to decode text descriptions into playable WAV audio files. You set generation parameters like duration and sampling controls to produce the desired music output.

Create sound effects from text descriptions for creative prototyping using the AudioGen workflow within Meta AudioCraft. It produces short sound effects like weather, crowds, or mechanical noises by applying neural decoding to your text prompts.

Melody-conditioned music generation produces music tracks guided by an existing melody, and AudioCraft supports this variant alongside standard text-to-music and stereo music generation workflows for flexible creative production.

Configure duration and sampling controls for audio generation by adjusting the runtime generation parameters in Meta AudioCraft. These settings define the length of the output and the sampling rate of the resulting WAV audio file.

Text-to-audio generation supports stereo output formats through specific MusicGen variants designed for stereo music generation. These variants process your text prompts to produce higher-channel playable audio tracks.

Neural audio codec reconstruction limitations depend on the EnCodec component, which encodes and decodes audio tokens to reconstruct waveforms. While it aims for higher fidelity, results vary based on the input tokens and sampling rate configurations.