RuntimeError: stft requires the return_complex parameter be given for real inputs about passt HOT 3 OPEN

Thank you!
It has been solved.
I have some additional questions.

I have several audio files of variable duration (40 secs, 1 min, etc.). Is it possible to use your library?
I am using the following:

import librosa

path = '/kaggle/input/audio-files/audio_files/S004.wav'
x , sr = librosa.load(path, sr=32000)

What can I do next?
Thank you again for your help!

from passt.

Hi!
Are you using the latest release ?
This issue should have been fixed in kkoutini/passt_hear21@dce8318

try uninstalling your current passt_hear21 package and reinstalling:

pip install -e 'git+https://github.com/kkoutini/[email protected]#egg=hear21passt' 

from passt.

Hi the model is compatible with the HEAR API.
Here is an example of the base model:

from hear21passt.base import load_model, get_scene_embeddings, get_timestamp_embeddings

model = load_model().cuda()
path = '/kaggle/input/audio-files/audio_files/S004.wav'
audio, sr = librosa.load(path, sr=32000)

embed, time_stamps = get_timestamp_embeddings(audio, model)
print(embed.shape)
embed = get_scene_embeddings(audio, model)
print(embed.shape)

If you need more control, take a look here where these methods are implemented.

from passt.