Improve preprocess.audio.load_wav() #8
Description
Is your feature request related to a problem? Please describe.
the load_wav() function is inefficient and can be improved for speed and performance.
Describe the solution you'd like
- Add condition to only resample when Original sr is different from target sr
- Avoid conversion from tensors to numpy arrays multiple times
- Check if this line is necessary: if len(tf.shape(y)) > 1:
Describe alternatives you've considered
Use time it to compare the performance of simply loading and resampling with librosa
Additional context
Add any other context or screenshots about the feature request here.