strip down file_name in dataset

nb_samtale.py

@@ -155,6 +155,7 @@ class NBSamtale(datasets.GeneratorBasedBuilder):
             for row in datalines:
                 data = json.loads(row)
                 audio_path = data["file_name"]
+                data["file_name"] = data["file_name"].split("/")[-1]
                 data["transcription"] = data[self.config.name]
                 meta[audio_path] = data