Switch to parquet

clustering_files.py

@@ -82,7 +82,7 @@ class ClusteringFiles(datasets.GeneratorBasedBuilder):
 
     def _generate_examples(self, file_path):
         """Yields examples."""
-        df = pd.read_pickle(file_path)
+        df = pd.read_parquet(file_path)
 
         files = {}
         for idx, row in df.iterrows():
@@ -93,9 +93,9 @@ class ClusteringFiles(datasets.GeneratorBasedBuilder):
                 }
             files[row["filename"]]["segments"].append({
                 "segment_id": row["segment_id"],
-                "speaker": row["speaker"],
+                "speaker": row.get("speaker", 'unknown'),
                 "duration": row["duration"],
-                "segment_clean": row["segment_clean"],
+                "segment_clean": row.get("segment_clean", True),
                 "start": row['start'],
                 "end": row['end'],
                 "readable_start": row['readable_start'],

dataset.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16963c709d10f4491adb04b8c2566cd158f6f3966eec34d223103f916f272036
+size 449338