def load_audio_classifier(folder):
print('Loading config folder: ' + folder)
config_file, caffemodel_file, net_proto = get_filenames(folder)
sbd.SbdConfig(config_file)
temp_proto = make_audio_temp_deploy(folder, net_proto)
net = caffe.Net(temp_proto, caffemodel_file, caffe.TEST)
classifier = AudioClassifier(net)
return classifier
def main(model_folder, example_folder):
config_file, caffemodel_file, net_proto = get_filenames(model_folder)
sbd.SbdConfig(config_file)
ctm_file, pitch_file, energy_file = get_audio_files(example_folder)
# parse ctm_file, pitch_file and energy_file
parser = AudioParser(ctm_file, pitch_file, energy_file)
parser.parse()
classifier = load_audio_classifier(model_folder)
data = classifier.predict_audio(parser)
print(data)
def load_lexical_classifier(folder, vector):
print('Loading config folder: ' + folder)
config_file, caffemodel_file, net_proto = get_filenames(folder)
sbd.SbdConfig(config_file)
temp_proto = make_lexical_temp_deploy(folder, net_proto)
net = caffe.Net(temp_proto, caffemodel_file, caffe.TEST)
if vector:
classifier = LexicalClassifier(net, vector)
else:
classifier = LexicalClassifier(net, vector)
return classifier
plain_text_instances_file.write(s.encode('utf8'))
# write to level db
level_db.write_training_instance(training_instance)
plain_text_instances_file.close()
if __name__ == '__main__':
parser = argparse.ArgumentParser(
description='create test and train datasets as a lmdb.')
parser.add_argument('config_file', help="path to config file")
args = parser.parse_args()
# initialize config
sbd.SbdConfig(args.config_file)
# create proper name for the database
SENTENCE_HOME = os.environ['SENTENCE_HOME']
data_folder = "/mnt/naruto/sentence/data/"
LEVEL_DB_DIR = "leveldbs"
database = SENTENCE_HOME + "/" + LEVEL_DB_DIR + "/" + sbd.SbdConfig.get_db_name_from_config(
sbd.config)
# check if database already exists
if os.path.isdir(database):
print("Deleting " + database + ". y/N?")
sys.stdout.flush()
s = raw_input()
if s != "Y" and s != "y":
def load_config(model_folder, model):
default_model = os.path.join(route_folder, model_folder, model)
config_file, caffemodel_file, net_proto = get_filenames(default_model)
sbd.SbdConfig(config_file)
parser.add_argument('vectorfile', help='the google news word vector', default='demo_data/GoogleNews-vectors-negative300.bin', nargs='?')
parser.add_argument('-nd','--no-debug', help='do not use debug mode, google vector is read', action='store_false', dest='debug', default=DEBUG)
args = parser.parse_args()
route_folder = args.routefolder
#### load lexical model ####
lexical_models = get_options(route_folder, LEXICAL_MODEL_FOLDER)
default_lexical_model = os.path.join(route_folder, LEXICAL_MODEL_FOLDER, lexical_models[0])
# get the caffe files
config_file, caffemodel_file, net_proto = get_filenames(default_lexical_model)
# read the config file
config_file = sbd.SbdConfig(config_file)
if not args.debug:
vector = Word2VecFile(args.vectorfile)
lexical_classifier = load_lexical_classifier(default_lexical_model, vector)
else:
vector = None
lexical_classifier = load_lexical_classifier(default_lexical_model, vector)
#### load audio model ####
audio_models = get_options(route_folder, AUDIO_MODEL_FOLDER)
default_audio_model = os.path.join(route_folder, AUDIO_MODEL_FOLDER, audio_models[0])
# get the caffe files
config_file, caffemodel_file, net_proto = get_filenames(default_audio_model)
def _load_config(self, model_folder):
config_file, caffemodel_file, net_proto = get_filenames(model_folder)
sbd.SbdConfig(config_file)