HViteOutputParse.py

#!/bin/env python

# Given threee files:
# token_timings (input) - containing:  start_t stop_t "token"
# master label file (input) - label file generated by HTK with class information
# output file
#
# merge the two input files to create a new timing file with the class label

# standard libraries
import os
import sys
import re

import pdb

# custom libraries
sys.path.append("c:/bin-htk/lib")
sys.path.append(os.path.expanduser("~/bin-htk/lib"))

import labels

def main():

    if len(sys.argv) != 4:
        raise ValueError, "Need token file and master label file"

    token_h = open(sys.argv[1])
    MLFs = labels.Labels()
    MLFs.add_mlf(sys.argv[2])
    output_h = open(sys.argv[3], 'w')
    
    line_match = re.compile('(?P<Start>\d+\.?\d*)+ (?P<Stop>[\d\.])+ "(?P<Token>.*)"')
    for line in token_h:
        match = line_match.match(line)
        if match:
            # Get information for current region
            (start, stop, token) = match.groups()
            # find class of token
            label_info = MLFs.access("%s.rec"%(token))
            label = label_info[0][0]    # pull out class label
            score = label_info[0][1]    # log likelihood of class

            # write out new label information
            output_h.write('%f %f "%s"\n'%(start, stop, label))

main()