Fix dataset so that it can be used in `load_dataset`

README.md

@@ -9,47 +9,8 @@ task_categories:
 - automatic-speech-recognition
 - audio-to-audio
 - audio-classification
-dataset_info:
-  features:
-  - name: audio
-    dtype: audio
-  - name: speaker_id
-    dtype: string
-  - name: transcript
-    dtype: string
-  - name: accent
-    dtype: string
-  - name: duration
-    dtype: float64
-  - name: age
-    dtype: string
-  - name: gender
-    dtype: string
-  splits:
-  - name: test
-    num_bytes: 496943021.995
-    num_examples: 5455
-  - name: val
-    num_bytes: 567088984.868
-    num_examples: 3779
-  - name: train
-    num_bytes: 81730701450.332
-    num_examples: 533436
-  download_size: 122944891314
-  dataset_size: 82794733457.195
-configs:
-- config_name: default
-  data_files:
-  - split: test
-    path: data/test-*
-  - split: val
-    path: data/val-*
-  - split: train
-    path: data/train-*
 ---
 
-
-
 # Important notice
 
 Differences between V2 version and [the version described in paper](https://huggingface.co/datasets/MushanW/GLOBE):