Skip to content

Preparing dataset without --interpolate fails #1

@maximyudayev

Description

@maximyudayev

Summary

Running this fails:

python prepare_dataset.py --sampling_rate 16000 --scale 4 --window_size 8192 --window_stride 4096 --batch_size 128 --low_pass --out_dir 'datasets'

I assume the error comes from here.

def add_data(h5_file, inputfiles, args, save_examples=False):

Expected behavior:

  1. Creates a folder whose name matches the arguments passed to the prepare_dataset.py.
  2. Produces h5 file with N patches of <window_size>, <window_stride> samples apart, interpreting WAV files as <sampling_rate> Hz audio, for each audio recording without interpolation.
  3. Expect low resolution patches to be smaller in window size than <window_size>, or to have less patches than in the full resolution h5 files.

Metadata

Metadata

Assignees

Labels

bugSomething isn't working

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions