set audio_column to audio instead of path

nst.py

@@ -91,7 +91,7 @@ class NSTDataset(datasets.GeneratorBasedBuilder):
             supervised_keys=None,
             homepage=_URL,
             task_templates=[
-                AutomaticSpeechRecognition(audio_column="path", transcription_column="text")
+                AutomaticSpeechRecognition(audio_column="audio", transcription_column="text")
             ],
         )