Follow the instructions here: https://github.com/tensorflow/models/tree/master/research/audioset/vggish

Specifically note the part about downloading the VGGish model checkpoint and Embedding PCA parameters. Be sure to run python vggish_smoke_test.py to ensure there are no errors with your VGGish setup.

To run live inference:

• python
• from live_inference import main as m
• m()

Notes:

Make sure to get the pyaudio installation correct.

pyaudio doesn't seemt to work on WSL2