Update multilingual-TEDX-fr.py

multilingual-TEDX-fr.py

@@ -88,7 +88,7 @@ class TEDX(datasets.GeneratorBasedBuilder):
 
     def _split_by_audio_file(self, segments_path: str, sentences_path: str, split_name: str) -> Tuple[List[str], List[List[Utterance]]]:
         speaker_paths = []
-        seen_ids = {}
+        seen_ids = set()
         segments = []
         with open(segments_path, "r") as segments, open(sentences_path) as sentences:
             segments_reader = csv.DictReader(segments, delimiter=' ', fieldnames=["segment_id", "speaker_id", "start_timestamp", "end_timestamp"])