Merge pull request #205 from prashanthellina/v3-fix-diarization

Author: m-bain

whisperx/alignment.py

@@ -450,8 +450,8 @@ def get_raw_text(word_row):
                 "end": srow["end"],
                 "text": text,
                 "words": word_list,
-                # "word-segments": wseg,
-                # "char-segments": cseg
+                "word-segments": wseg,
+                "char-segments": cseg
             }
         )
 

whisperx/transcribe.py

@@ -203,6 +203,12 @@ def cli():
 
     # >> Write
     for result, audio_path in results:
+        # Remove pandas dataframes from result so that
+        # we can serialize the result with json
+        for seg in result["segments"]:
+            seg.pop("word-segments", None)
+            seg.pop("char-segments", None)
+
         writer(result, audio_path, writer_args)
 
 if __name__ == "__main__":