def go(input_filename, mapping_dir, output_xml_filename, output_fsg_filename,
output_dict_filename, unit, word_unit, out_orth):
xml = load_xml(input_filename)
xml, fsg, dct = end_to_end(xml, input_filename, unit, word_unit, out_orth,
mapping_dir)
save_xml(output_xml_filename, xml)
save_txt(output_fsg_filename, fsg)
save_txt(output_dict_filename, dct)
def main(argv=None):
"""Hey! This function is named main!"""
parser = make_argparse()
args = parser.parse_args(argv)
if args.debug:
logging.basicConfig(level=logging.DEBUG)
else:
logging.basicConfig(level=logging.INFO)
if args.text_input:
tempfile, args.inputfile \
= create_input_xml(args.inputfile,
text_language=args.text_language,
save_temps=(args.outputfile
if args.save_temps else None))
if args.output_xhtml:
tokenized_xml_path = '%s.xhtml' % args.outputfile
else:
_, input_ext = os.path.splitext(args.inputfile)
tokenized_xml_path = '%s%s' % (args.outputfile, input_ext)
if os.path.exists(tokenized_xml_path) and not args.force_overwrite:
parser.error("Output file %s exists already, did you mean to do that?"
% tokenized_xml_path)
smil_path = args.outputfile + '.smil'
if os.path.exists(smil_path) and not args.force_overwrite:
parser.error("Output file %s exists already, did you mean to do that?"
% smil_path)
_, wav_ext = os.path.splitext(args.wavfile)
wav_path = args.outputfile + wav_ext
if os.path.exists(wav_path) and not args.force_overwrite:
parser.error("Output file %s exists already, did you mean to do that?"
% wav_path)
results = align_audio(args.inputfile, args.wavfile,
save_temps=(args.outputfile
if args.save_temps else None))
if args.output_xhtml:
convert_to_xhtml(results['tokenized'])
save_xml(tokenized_xml_path, results['tokenized'])
smil = make_smil(os.path.basename(tokenized_xml_path),
os.path.basename(wav_path), results)
shutil.copy(args.wavfile, wav_path)
save_txt(smil_path, smil)
def main(input_xml_path, input_smil_path, output_smil_path):
xml = load_xml(input_xml_path)
xml_filename = os.path.basename(input_xml_path)
smil = load_xml(input_smil_path)
ids = defaultdict(list)
for par in xpath_default(smil, ".//i:par"):
id = ""
for text_src in xpath_default(par, ".//i:text/@src"):
filename, id = text_src.split("#", 1)
filename = os.path.basename(filename)
if filename != xml_filename:
continue
for audio in xpath_default(par, ".//i:audio"):
filename = audio.attrib["src"]
begin = float(audio.attrib["clipBegin"])
end = float(audio.attrib["clipEnd"])
if not id:
continue
ids[id].append((filename, begin, end))
results = {"sentences": []}
for sentence in xpath_default(xml, ".//i:s"):
beginnings = defaultdict(lambda: 100000000000000.0)
endings = defaultdict(lambda: -1.0)
beginnings, endings = iterate_over_children(sentence, ids, beginnings,
endings)
for audio_path, beginning in beginnings.items():
results["sentences"].append({
"text_path": xml_filename,
"text_id": sentence.attrib["id"],
"audio_path": audio_path,
"start": beginning,
"end": endings[audio_path],
})
output_smil_text = pystache.render(SMIL_TEMPLATE, results)
save_txt(output_smil_path, output_smil_text)
def go(input_filename, output_filename, unit):
xml = load_xml(input_filename)
dct = make_dict(xml, input_filename, unit)
save_txt(output_filename, dct)
def go(seg_path, text_path, audio_path, output_path):
results = make_smil(text_path, audio_path, parse_hypseg(seg_path))
save_txt(output_path, results)
def go(input_filename, output_filename, unit):
xml = load_xml(input_filename)
fsg = make_fsg(xml, input_filename, unit)
save_txt(output_filename, fsg)
def save_txt_to_dir(output_path, dest_path, txt):
"""Save text to a directory, mimicking the interface of
save_txt_zip."""
save_txt(os.path.join(output_path, dest_path), txt)
def go(input_filename, output_filename, unit):
xml = load_xml(input_filename)
jsgf = make_jsgf(xml, input_filename, unit)
save_txt(output_filename, jsgf)