Encoding error in Python 3

[Juan Luis Cano Rodríguez]

I managed to train my own model with talon, but when trying to use it to extract a signature I got an encoding error:

ERROR when extracting signature with classifiers
Traceback (most recent call last):
  File "/home/jlcano/.miniconda3/envs/py3/lib/python3.5/site-packages/talon/signature/extraction.py", line 54, in extract
    markers = _mark_lines(lines, sender)
  File "/home/jlcano/.miniconda3/envs/py3/lib/python3.5/site-packages/talon/signature/extraction.py", line 81, in _mark_lines
    candidate = get_signature_candidate(lines)
  File "/home/jlcano/.miniconda3/envs/py3/lib/python3.5/site-packages/talon/signature/bruteforce.py", line 142, in get_signature_candidate
    markers = _mark_candidate_indexes(lines, candidate)
  File "/home/jlcano/.miniconda3/envs/py3/lib/python3.5/site-packages/talon/signature/bruteforce.py", line 166, in _mark_candidate_indexes
    markers = bytearray('c'*len(candidate))
TypeError: string argument without an encoding

[Minas Abrahamyan]

This is due talon code not ready for Python 3, this error occures with present built-in ML models too

[Sergey Obukhov]

Duplicate of #42