Failed to generate features: Failed to invoked DSP server: Unexpected token N in JSON at position 32


I have trained the model with accelerometer values. I am getting this error while i do live classification or model testing. Please find the attachment

Below are my findings

  1. I have trained the values similar to that of values collected from mobile that are scanned through qr code, by maintaining the window size and matching frequency.
  2. If i do live classification similar to that of training data that i have trained, it is working, but if i add data by maintaining the freq and time as did for training from my phone using the qr code, it is not working.
  3. I see the training dataset and testing dataset(collected from phone) are almost similar, and could not figure out the exact reason why it is throwing that error

Please let me know, where am i going wrong.


Hi @ajithkjajith,

The error seems to be related to wrong sensors’ names.
The Mobile Client uses the following sensors’ names on the accelerometer: accX, accY, accZ.
Your training dataset should also use those values for the sensors’ axis.


Thank you very much. That solved my issue.