HQSYN16

#!/usr/bin/python

# -*- coding: utf-8 -*-

import argparse

from asflight import AsfLight

# --------------------

def main():

    # all consonants

    phn_cons = [ 'b','c','C','d','D','w','W','f','g','h','x','G','j','k','l','L','m','M','H','n','J','N','p','r','P','R','Q','s','S','t','T','v','z','Z','!' ]

    # consonants without r, l, P, L

    phn_cons_m_left = [ 'b','c','C','d','D','w','W','f','g','h','x','G','j','k','m','M','H','n','J','N','p','R','Q','s','S','t','T','v','z','Z','!' ]

    # consonants without r, l, P, L, J

    phn_cons_m_right = [ 'b','c','C','d','D','w','W','f','g','h','x','G','j','k','m','M','H','n','N','p','R','Q','s','S','t','T','v','z','Z','!' ]

    parser = argparse.ArgumentParser( description = 'Convert ASF file to MLF.' )

    parser.add_argument( type=str, dest="asf_in", help="input ASF file", metavar="ASF_IN" )

    parser.add_argument( type=str, dest="asf_out", help="input ASF file", metavar="ASF_OUT" )

    parser.add_argument( "-c", "--code-page", type=str, dest="code_page",

                         help="encoding of all files, default value: %(default)s", default="utf-8" )

    # parse arguments

    args = parser.parse_args()

    asf = AsfLight( args.asf_in )

    for utt_name, asf_labels in asf.utts.iteritems():

        n_labels = len( asf_labels )

        for lab_idx in range( len( asf_labels ) ):

            curr_lab = asf_labels[ lab_idx ]

            prev_lab = asf_labels[ lab_idx-1 ] if ( lab_idx > 0 ) else None

            next_lab = asf_labels[ lab_idx+1 ] if ( lab_idx+1 < n_labels ) else None

            word_beg = ( curr_lab['word'] != '.' )

            word_end = ( next_lab is None ) or ( next_lab['word'] != '.' )

            if curr_lab['phone'] == 'n':

                if ( not word_end ) and ( next_lab['phone'] in [ 'g', 'k' ] ):

                    curr_lab['phone'] = 'N'

            elif curr_lab['phone'] == 'm':

                if ( not word_beg ) and ( prev_lab['phone'] in phn_cons_m_left ) and ( word_end or ( next_lab['phone'] in phn_cons_m_right ) ):

                    curr_lab['phone'] = 'H'

                elif ( not word_end ) and ( next_lab['phone'] in [ 'f', 'v' ] ):

                    curr_lab['phone'] = 'M'

            elif curr_lab['phone'] == 'x':

                if next_lab and ( next_lab['phone'] in [ 'b','d','D','w','W','g','h','G','z','Z' ] ):  # voiced paired consonants without 'v'

                    curr_lab['phone'] = 'G'

            elif curr_lab['phone'] == 'l':

                if ( not word_beg ) and ( prev_lab['phone'] in phn_cons ) and ( word_end or ( next_lab['phone'] in phn_cons ) ):

                    curr_lab['phone'] = 'L'

            elif curr_lab['phone'] == 'r':

                if ( not word_beg ) and ( prev_lab['phone'] in phn_cons ) and ( word_end or ( next_lab['phone'] in phn_cons ) ):

                    curr_lab['phone'] = 'P'

    asf.write( args.asf_out )

#

# ----------------------------------

#

# run the main program

if ( __name__ == "__main__" ):

    main()