test3

fleurs.py

@@ -52,7 +52,7 @@ for langs in _FLEURS_LANG:
 _ALL_CONFIGS.append("all")
 
 # TODO(FLEURS)
-_DESCRIPTION = "2.FLEURS is the speech version of the FLORES machine translation benchmark, covering 2000 n-way parallel sentences in n=102 languages."
+_DESCRIPTION = "3.FLEURS is the speech version of the FLORES machine translation benchmark, covering 2000 n-way parallel sentences in n=102 languages."
 _CITATION = ""
 _HOMEPAGE_URL = ""
 
@@ -104,7 +104,7 @@ class Fleurs(datasets.GeneratorBasedBuilder):
                 # "audio": datasets.Audio(sampling_rate=48_000),
                 "audio": datasets.features.Audio(sampling_rate=48_000),
                 "transcription": datasets.Value("string"),
-                "sentence": datasets.Value("string"),
+                "raw_transcription": datasets.Value("string"),
                 "lang_id": datasets.ClassLabel(names=langs),
                 "language": datasets.Value("string"),
                 "lang_group_id": datasets.ClassLabel(
@@ -204,7 +204,7 @@ class Fleurs(datasets.GeneratorBasedBuilder):
 
             data[file_name] = {
                 "id": int(_id),
-                "sentence": raw_transcription,
+                "raw_transcription": raw_transcription,
                 "transcription": transcription,
                 "num_samples": int(num_samples),
                 "lang_id": _FLEURS_LANG.index(lang_id),