def load_core_dev_files(timit):
core = set([l.strip().lower() for l in open("core_test")])
return [f for f in timit.utterances()
if f.split("-")[1].split("/")[0] not in core
and "sa1" not in f and "sa2" not in f][:200]
def load_brugnara_files(timit):
brugnara = set([l.strip() for l in open("corpus")])
return [f for f in timit.utterances()
if f.split("-")[1].split("/")[0] in brugnara
and "sa1" not in f and "sa2" not in f]
def load_training_files(timit):
#brugnara = set([l.strip() for l in open("corpus")])
return [f for f in timit.utterances()
if "sa1" not in f and "sa2" not in f]
engine.load(df)
# Compute the set of all phoneme types.
phoneme_map = {}
phoneme_set = speech.PhonemeSet()
for i, p in enumerate(set(timit.phones())):
phoneme = phoneme_set.phonemes.add()
phoneme.id = i
phoneme.name = p
phoneme_map[p] = i
print i, p
# All the male utterances of a region.
brugnara = set([l.strip() for l in open("corpus")])
#f.startswith("dr1-f")
utterance_names = [f for f in timit.utterances() if f.split("-")[1].split("/")[0] in brugnara and "sa1" not in f and "sa2" not in f]
print len(utterance_names)
utterance_set = speech.UtteranceSet()
all_features = []
#for utterance_file in utterance_names:
for utterance_file in utterance_names:
#for utterance_file in [ u for u in utterance_names if u == "dr8-mbcg0/sx57"]:
# extract features from an audio file using AudioFileProcessor
afp = AudioFileProcessor()
afp.processFile(engine, timit.abspath(utterance_file + ".wav"))
phone_times = timit.phone_times(utterance_file)
print phone_times
last = float(phone_times[-1][2])
features = engine.readAllOutputs()
