def lex_gen_ipa(locale, engine, voice, speak=False):
global tts
if engine == 'sequitur':
ipas = sequitur_gen_ipa(SEQUITUR_MODEL, lex_base)
else:
tts.locale = locale
tts.engine = engine
tts.voice = voice
ipas = tts.gen_ipa(lex_base)
if speak:
say_ipa('de', 'mary', 'dfki-pavoque-neutral-hsmm', ipas)
return ipas
else:
num_ts_complete += 1
cnt = 0
for item in reversed(sorted(missing.items(), key=lambda x: x[1])):
cnt += 1
if cnt > options.num_words:
break
if verbose:
logging.info(u"Missing %4d times: %s" % (item[1], item[0]))
else:
logging.info(item[0].encode('utf8'))
if options.generate:
ipas = sequitur_gen_ipa(sequitur_model, item[0])
logging.info(u"%4d/%4d generated lex entry: %s -> %s" %
(cnt, options.num_words, item[0], ipas))
lex[item[0]] = {'ipa': ipas}
logging.info(
"%d missing words total. %d submissions lack at least one word, %d are covered fully by the lexicon."
% (len(missing), num_ts_lacking, num_ts_complete))
if options.generate:
logging.info('saving lexicon...')
lex.save()
logging.info('saving lexicon...done.')
def lex_edit(token):
global lex, lex_token, lex_entry, lex_base
lex_set_token (token)
while True:
try:
lex_paint_main()
c = stdscr.getch()
if c == ord('q'):
lex.save()
break
# generate de-mary
elif c == ord('g'):
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
ipas = tts.gen_ipa (lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate de-espeak
elif c == ord('h'):
tts.locale ='de'
tts.engine ='espeak'
tts.voice ='de'
ipas = tts.gen_ipa (lex_base)
lex_entry['ipa'] = ipas
tts.locale ='de'
tts.engine ='mary'
tts.voice ='bits3'
tts.say_ipa(ipas)
# generate en-mary
elif c == ord('l'):
tts.locale ='en-US'
tts.engine ='mary'
tts.voice ='cmu-rms-hsmm'
ipas = tts.gen_ipa (lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate fr-mary
elif c == ord('k'):
tts.locale ='fr'
tts.engine ='mary'
tts.voice ='upmc-pierre-hsmm'
ipas = tts.gen_ipa (lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate de-sequitur
elif c == ord('j'):
ipas = sequitur_gen_ipa (SEQUITUR_MODEL, lex_base)
tts.locale ='de'
tts.engine ='mary'
tts.voice ='bits3'
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# speak de mary unitsel
elif c == ord('p'):
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
tts.say_ipa(ipas)
# speak de mary hsmm
elif c == ord('o'):
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'dfki-pavoque-neutral-hsmm'
tts.say_ipa(ipas)
# speak fr mary hsmm
elif c == ord('i'):
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'fr'
tts.engine = 'mary'
tts.voice = 'pierre-voice-hsmm'
tts.say_ipa(ipas)
# speak en mary hsmm
elif c == ord('u'):
ipas = lex_entry['ipa']
tts.locale = 'en-US'
tts.engine = 'mary'
tts.voice = 'cmu-rms-hsmm'
tts.say_ipa(ipas)
# edit token
elif c == ord('t'):
token = misc.edit_popup(stdscr, ' Token ', '')
lex_set_token (token)
# edit XS
elif c == ord('e'):
ipas = lex_entry['ipa']
xs = ipa2xsampa (lex_token, ipas, stress_to_vowels=False)
xs = misc.edit_popup(stdscr, ' X-SAMPA ', xs)
ipas = xsampa2ipa (lex_token, xs)
lex_entry['ipa'] = ipas
except:
logging.error('EXCEPTION CAUGHT %s' % traceback.format_exc())
def lex_edit(token):
global lex, lex_token, lex_entry, lex_base
lex_set_token(token)
while True:
try:
lex_paint_main()
c = raw_input('lex > ').lower()
if c == 'q':
lex.save()
break
# generate de-mary
elif c == 'g':
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
ipas = tts.gen_ipa(lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate de-espeak
elif c == 'h':
tts.locale = 'de'
tts.engine = 'espeak'
tts.voice = 'de'
ipas = tts.gen_ipa(lex_base)
lex_entry['ipa'] = ipas
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
tts.say_ipa(ipas)
# generate en-mary
elif c == 'l':
tts.locale = 'en-US'
tts.engine = 'mary'
tts.voice = 'cmu-rms-hsmm'
ipas = tts.gen_ipa(lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate fr-mary
elif c == 'k':
tts.locale = 'fr'
tts.engine = 'mary'
tts.voice = 'upmc-pierre-hsmm'
ipas = tts.gen_ipa(lex_base)
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# generate de-sequitur
elif c == 'j':
ipas = sequitur_gen_ipa(SEQUITUR_MODEL, lex_base)
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
tts.say_ipa(ipas)
lex_entry['ipa'] = ipas
# speak de mary unitsel
elif c == 'p':
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'bits3'
tts.say_ipa(ipas)
# speak de mary hsmm
elif c == 'o':
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'de'
tts.engine = 'mary'
tts.voice = 'dfki-pavoque-neutral-hsmm'
tts.say_ipa(ipas)
# speak fr mary hsmm
elif c == 'i':
if len(lex_entry['ipa']) == 0:
continue
ipas = lex_entry['ipa']
tts.locale = 'fr'
tts.engine = 'mary'
tts.voice = 'pierre-voice-hsmm'
tts.say_ipa(ipas)
# speak en mary hsmm
elif c == 'u':
ipas = lex_entry['ipa']
tts.locale = 'en-US'
tts.engine = 'mary'
tts.voice = 'cmu-rms-hsmm'
tts.say_ipa(ipas)
# edit token
elif c == 't':
readline.add_history(lex_token.encode('utf8'))
token = raw_input('token: ').decode('utf8')
lex_set_token(token)
# edit XS
elif c == 'e':
ipas = lex_entry['ipa']
xs = ipa2xsampa(lex_token, ipas, stress_to_vowels=False)
readline.add_history(xs.encode('utf8'))
xs = raw_input('X-SAMPA: ').decode('utf8')
ipas = xsampa2ipa(lex_token, xs)
lex_entry['ipa'] = ipas
except:
logging.error('EXCEPTION CAUGHT %s' % traceback.format_exc())
for word in tokenize(ts['prompt']):
if word in lex:
continue
if word in missing:
missing[word] += 1
else:
missing[word] = 1
cnt = 0
for item in reversed(sorted(missing.items(), key=lambda x: x[1])):
lex_base = item[0]
ipas = sequitur_gen_ipa (lex_base)
logging.info ( u"%5d/%5d Adding missing word : %s [ %s ]" % (cnt, len(missing), item[0], ipas) )
lex_entry = {'ipa': ipas}
lex[lex_base] = lex_entry
cnt += 1
#
# dictionary export
#
dictfn2 = '%s/local/dict/lexicon.txt' % data_dir
for word in tokenize(ts['prompt']):
if word in lex:
continue
if word in missing:
missing[word] += 1
else:
missing[word] = 1
cnt = 0
for item in reversed(sorted(missing.items(), key=lambda x: x[1])):
lex_base = item[0]
ipas = sequitur_gen_ipa(SEQUITUR_MODEL % options.lang, lex_base)
logging.info(u"%5d/%5d Adding missing word : %s [ %s ]" %
(cnt, len(missing), item[0], ipas))
lex_entry = {'ipa': ipas}
lex[lex_base] = lex_entry
cnt += 1
#
# dictionary export
#
dictfn2 = '%s/local/dict/lexicon.txt' % data_dir
logging.info("Exporting dictionary...")
