parser.add_argument('-mlfname', type=str, help="The name to prepend the output mlfs, if making mlfs.", default="txt")
parser.add_argument('-slf', action="store_true", help="Output slfs.")
parser.add_argument('-pronoun_variant', action="store_true", help="Create pronounciation variant slfs.")
parser.add_argument('combilexpath', type=str, help="The path to the combilex dictionary directory.")
args = parser.parse_args()
#Used for utt creation
args.intype = "txt"
args.stanfordparse = False
args.dictionary = dictionary.Dictionary(args.combilexpath)
if args.pronoun_variant and not args.slf:
print "Cannot create pronounciation variant mlfs. Please output slfs."
sys.exit()
txtfiles = io.load_txt_dir(args.txtdir)
#Opening the mlf files here means we don't have to loop twice if outputting slfs as well.
if args.mlf:
#Out mlf with short pause
wfsp = open(os.path.join(args.outdir, args.mlfname+"_sp.mlf"), "w")
wfsp.write("#!MLF!#\n")
#out mlf without short pause
wfnosp = open(os.path.join(args.outdir, args.mlfname+"_no_sp.mlf"), "w")
wfnosp.write("#!MLF!#\n")
for txt in txtfiles:
print "Processing {0}".format(txt)
#Make an utt
utt = utterance.Utterance(txt, args)
if args.mlf:
parser.add_argument('-qpath', type=str, help="The directory to write the question set to.", default=os.path.join("questions", str(datetime.now())+".hed"))
parser.add_argument('-target', type=str, help="The target type of the output labs and questions.", choices=['HMM', 'NN'], default='HMM')
parser.add_argument('-stanfordparse', action="store_true", help="Add stanford parse information from parses in provided dirpath. Note this assumes you have already run txt2parse to create a parse.")
parser.add_argument('-context_type', type=str, choices=['absolute', 'relational'], help="The type of positional contexts to add.", default='relational')
parser.add_argument('-parsedir', type=str, help="The path to the parses.", default="parse")
parser.add_argument('-HHEd_fix', action="store_true", help="Applies a fix to the contexts around the current phoneme to be compatible with hardcoded delimiters in HHEd.")
args = parser.parse_args()
#The phoneme set used - hardcoded as currently only combilex is possible
args.phoneme_features = phoneme_features.CombilexPhonemes()
if args.stanfordparse:
args.parsedict = read_stanford_parses(args.parsedir)
if args.intype == "txt":
labs = io.load_txt_dir(args.txtdir)
if args.combilexpath == None:
print "No path to combilex. Please use -combilexpath option."
sys.exit()
args.dictionary = dictionary.Dictionary(args.combilexpath)
elif args.intype == "hts_lab":
labs = io.open_line_by_line(args.inpath)
args.intype = "hts_mlf"
else:
if args.inpath == None:
print "Input is mlf type but no mlf path has been set! Please use -inpath option."
sys.exit()
mlf = open(args.inpath, "r").readlines()
labs = parse_mlf(mlf, args.intype)
#Used if we make questions fitted to a dataset