mila-udem /
fuel
| Total Complexity | 1 |
| Total Lines | 23 |
| Duplicated Lines | 0 % |
| 1 | from fuel.datasets.hdf5 import H5PYDataset |
||
| 5 | class YouTubeAudio(H5PYDataset): |
||
| 6 | """Dataset of audio from YouTube video. |
||
|
|
|||
| 7 | |||
| 8 | Assumes the existence of a dataset file with the name |
||
| 9 | `youtube_id.hdf5`. These datasets don't have any split; the entire |
||
| 10 | audio sequence is considered training. |
||
| 11 | |||
| 12 | Note that the data structured in the form `(batch, time, features)` |
||
| 13 | where `features` are the audio channels (dimension 1 or 2) and batch is |
||
| 14 | equal to 1 in this case (since there is only one audiotrack). |
||
| 15 | |||
| 16 | Parameters |
||
| 17 | ---------- |
||
| 18 | youtube_id : str |
||
| 19 | 11-character video ID (taken from YouTube URL) |
||
| 20 | \*\*kwargs |
||
| 21 | Passed to the `H5PYDataset` class. |
||
| 22 | |||
| 23 | """ |
||
| 24 | def __init__(self, youtube_id, **kwargs): |
||
| 25 | super(YouTubeAudio, self).__init__( |
||
| 26 | file_or_path=find_in_data_path('{}.hdf5'.format(youtube_id)), |
||
| 27 | which_sets=('train',), **kwargs |
||
| 28 | ) |
||
| 29 |
Escape sequences in Python are generally interpreted according to rules similar to standard C. Only if strings are prefixed with
rorRare they interpreted as regular expressions.The escape sequence that was used indicates that you might have intended to write a regular expression.
Learn more about the available escape sequences. in the Python documentation.