Add segmentation in retrieval scripts #21

tyiannak / deep_audio_features

Pytorch implementation of deep audio embedding calculation

MIT License

98 stars 9 forks source link

Closed tyiannak closed 3 years ago

tyiannak commented 3 years ago

@lobracost please do a quick QA. Instructions:

use 10-20 songs as a db collection (say stored in collection_path)
use 4 cnn pertained models (say stored in models_path)
build the db: python3 bin/deep_retrieval_build_db.py -m models_path -i collection_path
run some example queries (u can even use songs from the db for sanity check): python3 bin/deep_retrieval_query.py -d db -I some_song.wav

pakoromilas commented 3 years ago

The functionality works as expected on my machine, while the results seem reasonable.