Use the dataset viewer for audio/text previews

README.md

@@ -13,7 +13,7 @@ annotations_creators:
 - machine-generated
 multilinguality: multilingual
 source_datasets:
-- derived-from-ZeroAgency/shkolkovo-bobr.video-webinars-audio
+- ZeroAgency/shkolkovo-bobr.video-webinars-audio
 tags:
 - audio
 - russian
@@ -28,6 +28,34 @@ configs:
   data_files:
   - split: train
     path: data/train-*.parquet
+dataset_info:
+  features:
+  - name: audio
+    dtype: audio
+  - name: text
+    dtype: string
+  - name: id
+    dtype: string
+  - name: duration
+    dtype: float64
+  - name: language
+    dtype: string
+  - name: recording_id
+    dtype: string
+  - name: sample_rate
+    dtype: int32
+  - name: start_seconds
+    dtype: float64
+  - name: end_seconds
+    dtype: float64
+  - name: source_spans
+    list:
+    - name: start_sample
+      dtype: int64
+    - name: end_sample
+      dtype: int64
+  - name: removed_pause_seconds
+    dtype: float64
 ---
 
 # Webinars