| import datasets |
| import os |
| import shutil |
| |
| train0, validation0, test0 = datasets.load_dataset("superb", "ks", split=["train","validation","test"]) |
|
|
| labels = train0.features["label"].names |
| label2id = {x: labels.index(x) for x in labels} |
| |
|
|
| down_id = label2id['down'] |
| on_id = label2id['on'] |
|
|
| train0_down = train0.filter(lambda example: example['label'] == down_id) |
| train0_on = train0.filter(lambda example: example['label'] == on_id) |
|
|
| shutil.rmtree('/home/mr249/ac_h/down_on/data/train/down',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/train/down') |
| shutil.rmtree('/home/mr249/ac_h/down_on/data/train/on',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/train/on') |
|
|
| for e in train0_down: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/train/down/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
| for e in train0_on: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/train/on/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
| validation0_down = validation0.filter(lambda example: example['label'] == down_id) |
| validation0_on = validation0.filter(lambda example: example['label'] == on_id) |
|
|
| shutil.rmtree('/home/mr249/ac_h/down_on/data/validation/down',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/validation/down') |
| shutil.rmtree('/home/mr249/ac_h/down_on/data/validation/on',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/validation/on') |
|
|
| for e in validation0_down: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/validation/down/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
| for e in validation0_on: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/validation/on/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
| test0_down = test0.filter(lambda example: example['label'] == down_id) |
| test0_on = test0.filter(lambda example: example['label'] == on_id) |
|
|
| shutil.rmtree('/home/mr249/ac_h/down_on/data/test/down',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/test/down') |
| shutil.rmtree('/home/mr249/ac_h/down_on/data/test/on',ignore_errors=True) |
| os.mkdir('/home/mr249/ac_h/down_on/data/test/on') |
|
|
| for e in test0_down: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/test/down/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
| for e in test0_on: |
| p = e['audio']['path'] |
| newpath = '/home/mr249/ac_h/down_on/data/test/on/{}'.format(p.split('/')[len(p.split('/')) - 1]) |
| os.link(p,newpath) |
|
|
|
|