113 lines
2.7 KiB
Python
113 lines
2.7 KiB
Python
#
|
|
# forced alignment using novo-api.
|
|
#
|
|
# *** IMPORTANT ***
|
|
# This file should be treated as confidencial.
|
|
# This file should not be copied or uploaded to public sites.
|
|
#
|
|
# NOTES:
|
|
# The usage of novo api: https://bitbucket.org/novolanguage/python-novo-api
|
|
# I couldn't make it work as I described in the mail to Martijn Bartelds on
|
|
# 2018/12/03.
|
|
# As per the advice from him, I modified testgrammer.py and made it a function.
|
|
#
|
|
# In order to run on Python 3.6, the following points are changed in novo-api.
|
|
# (1) backend/__init__.py
|
|
# - #import session
|
|
# from . import session
|
|
# (2) backend/session.py
|
|
# - #except Exception, e:
|
|
# except Exception as e:
|
|
# - #print self.last_message
|
|
# print(self.last_message)
|
|
# (3) asr/segment/praat.py
|
|
# - def print_tier(output, title, begin, end, segs, (format, formatter))
|
|
# def print_tier(output, title, begin, end, segs, format, formatter):
|
|
# (4) asr/spraaklab/__init.py
|
|
# - #import session
|
|
# from . import session
|
|
# (5) asr/spraaklab/schema.py
|
|
# - #print data, "validated not OK", e.message
|
|
# print("{0} validated not OK {1}".format(data, e.message))
|
|
# - #print data, "validated OK"
|
|
# print("{} validated OK".format(data))
|
|
# - #if isinstance(object, basestring):
|
|
# if isinstance(object, str)
|
|
#
|
|
# Aki Kunikoshi
|
|
# 428968@gmail.com
|
|
#
|
|
import argparse
|
|
import json
|
|
|
|
from novoapi.backend import session
|
|
|
|
# username / password cannot be passed as artuments...
|
|
p = argparse.ArgumentParser()
|
|
#p.add_argument("--user", default=None)
|
|
#p.add_argument("--password", default=None)
|
|
p.add_argument("--user", default='martijn.wieling')
|
|
p.add_argument("--password", default='fa0Thaic')
|
|
args = p.parse_args()
|
|
|
|
wav_file = 'c:\\OneDrive\\WSL\\test\\onetwothree.wav'
|
|
|
|
rec = session.Recognizer(grammar_version="1.0", lang="nl", snodeid=101, user=args.user, password=args.password, keepopen=True) # , modeldir=modeldir)
|
|
grammar = {
|
|
"type": "confusion_network",
|
|
"version": "1.0",
|
|
"data": {
|
|
"kind": "sequence",
|
|
"elements": [{
|
|
"kind": "word",
|
|
"pronunciation": [{
|
|
"phones": ["wv",
|
|
"a1",
|
|
"n"],
|
|
"id": 0
|
|
},
|
|
{
|
|
"phones": ["wv",
|
|
"uh1",
|
|
"n"],
|
|
"id": 1
|
|
}],
|
|
"label": "one"
|
|
},
|
|
{
|
|
"kind": "word",
|
|
"pronunciation": [{
|
|
"phones": ["t",
|
|
"uw1"],
|
|
"id": 0
|
|
}],
|
|
"label": "two"
|
|
},
|
|
{
|
|
"kind": "word",
|
|
"pronunciation": [{
|
|
"phones": ["t",
|
|
"r",
|
|
"iy1"],
|
|
"id": 0
|
|
},
|
|
{
|
|
"phones": ["s",
|
|
"r",
|
|
"iy1"],
|
|
"id": 1
|
|
}],
|
|
"label": "three"
|
|
}]
|
|
},
|
|
"return_objects": ["grammar"],
|
|
"phoneset": "novo70"
|
|
}
|
|
|
|
res = rec.setgrammar(grammar)
|
|
#print "Set grammar result", res
|
|
|
|
#res = rec.recognize_wav("test/onetwothree.wav")
|
|
res = rec.recognize_wav(wav_file)
|
|
#print "Recognition result:", json.dumps(res.export(), indent=4)
|