search

Strong-AI-Lab / emotion

Emotion Recognition ToolKit (ERTK): tools for emotion recognition. Dataset processing, feature extraction, experiments,
MIT License
55 stars 11 forks source link

Speaker in SAVEE dataset #10

Closed bagustris closed 11 months ago

bagustris commented 11 months ago

Should the following line use p.parts[1] instead of p.stem[:2] for obtaining speaker ID?

https://github.com/Strong-AI-Lab/emotion/blob/076ad6e6e55a573693daa4e0e3904c0f110e0c15/datasets/SAVEE/process.py#L57

I checked with these (p equal to wav_file in this case)

In [99]: wav_file.parts[1]
Out[99]: 'JK'

In [100]: wav_file.stem[2]
Out[100]: 'n2'
bagustris commented 11 months ago

I just checked that you sampled and renamed the original files, so this is not the issue.