🎸 Generating Music using Audiocraft and W&B 🐝

In this notebook we demonstrate how you can generate music and other types of audio from text prompts or generate new music from existing music using SoTA models such as MusicGen and AudioGen from Audiocraft and play and visualize them using Weights & Biases.

If you want to know more about the underlying architectures for MusicGen and AudioGen and explore some cool audio samples generated by these models, you can check out this W&B report.

This is how the W&B Table looks like with the interactive audio player, waveform visualizer and spectrogram visualization along with the prompts and other configs. Note that the notebook automatically sets the seed if you leave it blank, so your experiments are always reproducible.

If you want to know more about the underlying architectures for MusicGen and AudioGen and explore some cool audio samples generated by these models, you can check out this W&B report.