Notes

AudioLDM

Generates audio from words. Think Dall-E (or Craiyon or Midjourne) but for sound.

Written in python (but readme has only CLI examples). (To install: pip install audioldm)

huggingface hosts a GUI to try it out. I tried "" and got this sound (could only figure out how to download as video).

How can we use this? To generate and transform (therefore expand, enhance, etc.) targeted audio data. This can help when we don't have any, or not enough, data to get good models, or a good sense of how robust our models are.

Some links

• huggingface
• homepage
• github
• pypi
• video about AudioLDM
• paper

Links

• Generate audio with python
• Multipurpose Library for Synthetic Time Series