search

Audio-AGI / AudioSep

Official implementation of "Separate Anything You Describe"
https://audio-agi.github.io/Separate-Anything-You-Describe/
MIT License
1.63k stars 118 forks source link

ImportError: cannot import name 'inference' from 'pipeline' #41

Open jurassicjordan opened 6 months ago

jurassicjordan commented 6 months ago

this is probably just me being really bad at coding, I'm trying to run the example inference code in README and am getting this error: ImportError: cannot import name 'inference' from 'pipeline' (/home/jordancruz/Tools/AudioSep/pipeline.py)

am I doing something wrong?

atreydesai commented 5 months ago

They changed the name of the function and didn't update the README. Replace all mentions of inference with separate_audio in the example code and it will run.

clove-angel commented 3 months ago

Using python benchmark.py --checkpoint_path audiosep_base_4M_steps.ckpt displays an error FileNotFoundError: [Errno 2] No such file or directory: 'evaluation/data/clotho\segment-0.wav' How to use clotho in the evaluation data I downloaded through the link?

atreydesai commented 3 months ago

You have to download the clotho dataset here: https://paperswithcode.com/dataset/clotho