fix: remove empty files.

src/coral_models/prepare_raw_data.py

@@ -344,9 +344,16 @@ def prepare_raw_data(
     # audio.
     read_aloud_duration = 0.0
     conversation_duration = 0.0
+    rows_to_remove = []
     for row_i, row in tqdm(recordings.iterrows()):
         filename = input_path / row["filename"]
 
+        # Check if the file is empty, and if it is, remove it from the dataframe
+        # and continue to the next file
+        if filename.stat().st_size < 200000:  # Any file smaller than this is empty
+            rows_to_remove.append(row_i)
+            continue
+
         # Get the new filename
         # New filename is in the format is for conversations:
         # "recording_id_speaker_id1_speaker_id2_recorder_speaker_id_conversation.wav"
@@ -400,6 +407,9 @@ def prepare_raw_data(
         except FileNotFoundError:
             pass
 
+    # Remove rows with empty files
+    recordings = recordings.drop(rows_to_remove).reset_index(drop=True)
+
     # Write a README file
     readme = make_readme()
     with open(output_path / "README.md", "w") as f: