Update README.md

README.md

@@ -14,7 +14,7 @@ We preprocessed the Visual Accent and Dialect Archive (https://archive.mith.umd.
 This version currently only contains read speech, specifically, readings of the Rainbow Passage, which contains every sound in English.
 
 
-Citation:
+## Citation:
 ```
 @misc{vada,
   title     = {{Visual Accent and Dialect Archive}},
@@ -30,3 +30,47 @@ Citation:
   note      = {Accessed on DATE at \url{https://huggingface.co/datasets/Berkeley-NLP/visual_accent_dialect_archive}},
 }
 ```
+
+## Loading the dataset (example)
+```python
+from huggingface_hub import snapshot_download
+from datasets import load_dataset, Audio, Video
+
+repo = "Berkeley-NLP/visual_accent_dialect_archive"
+
+# Download dataset snapshot (cached by HF)
+root = snapshot_download(repo_id=repo, repo_type="dataset")
+
+# Load metadata CSV
+ds = load_dataset(
+    "csv",
+    data_files={"test": f"{root}/test.csv"},
+    split="test",
+)
+
+# Rename path columns
+ds = ds.rename_columns({
+    "audio_segment_path": "audio",
+    "video_noaudio_segment_path": "video_no_audio",
+    "video_withaudio_segment_path": "video_with_audio",
+})
+
+# Convert repo-relative paths to absolute local paths
+def absolutize(ex):
+    ex["audio"] = f"{root}/" + ex["audio"]
+    ex["video_no_audio"] = f"{root}/" + ex["video_no_audio"]
+    ex["video_with_audio"] = f"{root}/" + ex["video_with_audio"]
+    return ex
+
+ds = ds.map(absolutize)
+
+# Decode audio and video
+ds = ds.cast_column("audio", Audio())
+ds = ds.cast_column("video_no_audio", Video())
+ds = ds.cast_column("video_with_audio", Video())
+
+ex = ds[0]
+print(ex["audio"]["sampling_rate"], ex["audio"]["array"].shape)
+print(ex["video_no_audio"])
+print(ex["video_with_audio"])
+```