removed tokens array from transcripts

handler.py

@@ -48,9 +48,11 @@ class EndpointHandler():
         t0 = time.time()
         transcript = self.model.transcribe(path_to_audio, **decode_options)
         t1 = time.time()
-        
         total = t1-t0
         print(f'Finished transcription in {total} seconds')
+        for segment in transcript['segments']:
+          # Remove the tokens array, it was making response too verbose
+          segment.pop('tokens', None)
         
 
         # postprocess the prediction