make dictionary calls more flexible

audio-kw-in-context.py

@@ -132,7 +132,7 @@ class AudioKwInContext(datasets.GeneratorBasedBuilder):
                     "file": tfile,
                     "sentence": row['sentence'],
                     "language": self.config.language,
-                    "speaker_id": row['speaker_id'],
+                    "speaker_id": row.get('speaker_id',row.get('client_id')),
                     "keywords": row['keywords'],  
                     "audio": tfile,
                     "start_times": row.get('start_times'),