|
import os.path |
|
from glob import glob |
|
from os.path import join as p_join |
|
from datasets import Dataset, Audio, DatasetDict |
|
|
|
for root_dir in glob(p_join("audio", "*")): |
|
experiment_name = os.path.basename(root_dir) |
|
data_dict = {} |
|
for audio_dir in sorted(glob(p_join(root_dir, "*"))): |
|
data_dict[os.path.basename(audio_dir)] = sorted(glob(p_join(audio_dir, "*"))) |
|
|
|
dataset = Dataset.from_dict(data_dict) |
|
for key in data_dict: |
|
dataset = dataset.cast_column(key, Audio()) |
|
DatasetDict({"test": dataset}).push_to_hub( |
|
"asahi417/experiment-audio-tokenizer", config_name=experiment_name |
|
) |
|
|