0

I want to test out Tensorflow's audio recognition by training my own set of data. Following the instructions from the Tensorflow Simple Audio Recognition Tutorial:

https://www.tensorflow.org/tutorials/sequences/audio_recognition

I receive an error when I run the train.py: SyntaxErorr: 'unicodescape' codec can't decode byes in position 2-3...

I changed the --data_dir=/your/data/folder/ to a folder in my computer with my own set of data by like this in the train script:

parser.add_argument(
  '--data_dir',
  type=str,
  default='C:\Users\Matthew Chen\Documents\GSF\ALL',
  help="""\
  Where to download the speech training data to.
  """)

where ALL is the folder with all of my WAV files.

I assume that I have not disabled --data_url= correctly but the documentation does not seem to provide enough detail onto how to disable it.

This is what the --data_url= looks like:

parser.add_argument(
  '--data_url',
  type=str,
  pylint: disable=line-too-long
  #default='C:\Users\Matthew Chen\Documents\GSF\...\ALL',
  # pylint: enable=line-too-long
  help='Location of speech training data archive on the web.')

Is it possible that I have to input a specific string to disable it?

James Wang
  • 55
  • 5
  • have a look at this: https://stackoverflow.com/questions/1347791/unicode-error-unicodeescape-codec-cant-decode-bytes-cannot-open-text-file – lCapp Aug 06 '18 at 18:03
  • `python tensorflow/examples/speech_commands/train.py \ ` `--data_url= \ ` `--data_dir=/your/data/folder \ ` `--summaries_dir=/your/dir \ ` `--wanted_words=left,right,up,down` – BhanuKiran Oct 18 '18 at 12:01

0 Answers0