Update multilingual-TEDX-fr.py

multilingual-TEDX-fr.py

@@ -89,7 +89,7 @@ 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 = set()
-        segments = []
+        segments_by_speaker = []
         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"])
             sentences_list = sentences.readlines()
@@ -97,14 +97,14 @@ class TEDX(datasets.GeneratorBasedBuilder):
                 if segment["speaker_id"] not in seen_ids:
                     seen_ids.add(segment["speaker_id"])
                     speaker_paths.append(Path("data") / Path(split_name) / Path("wav") / Path(f"{segment['speaker_id']}.flac"))
-                    segments.append([])
-                segments[-1].append(Utterance(speaker_id=segment["speaker_id"],
+                    segments_by_speaker.append([])
+                segments_by_speaker[-1].append(Utterance(speaker_id=segment["speaker_id"],
                                               index=int(segment["segment_id"].split("_")[1]),
                                               sentence=sentence,
                                               start_timestamp=segment["segment_start_timestamp"],
                                               end_timestamp=segment["segment_end_timestamp"]
                                               ))
-        return speaker_paths, segments
+        return speaker_paths, segments_by_speaker