def main():
parser = argparse.ArgumentParser()
parser.add_argument("-g", "--detect_gender", type=bool, default=False, help="Enable gender detection")
parser.add_argument("-d", "--vad_engine", choices=['sm', 'smn'], default='smn', help="Voice activity detection: smn (default) or sm")
parser.add_argument("-b", "--ffmpeg_binary", default='ffmpeg', help="FFMPEG binary")
parser.add_argument("input", help="Input file")
parser.add_argument("output", help="Output file")
args = parser.parse_args()
seg = Segmenter(vad_engine=args.vad_engine, detect_gender=args.detect_gender, ffmpeg=args.ffmpeg_binary)
seg.batch_process([args.input], [args.output], verbose=True)
def run_inaseg(input_wav, csv_out_dir):
# load neural network into memory, may last few seconds
with warnings.catch_warnings():
warnings.simplefilter("ignore", category=MatplotlibDeprecationWarning)
from inaSpeechSegmenter import Segmenter
seg = Segmenter(vad_engine="smn", detect_gender=True)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
stem = input_wav.stem
input_files = [str(input_wav)]
output_files = [str(csv_out_dir / f"{stem}.csv")]
seg.batch_process(input_files, output_files, verbose=True)
#subprocess.run(["ina_speech_segmenter.py", "-i", input_wav, "-o", csv_out_dir])
return get_csv_path(input_wav, csv_out_dir)
def test_batch(self):
seg = Segmenter(vad_engine='sm')
with tempfile.TemporaryDirectory() as tmpdirname:
lout = [os.path.join(tmpdirname, '1.csv'), os.path.join(tmpdirname, '2.csv')]
ret = seg.batch_process(['./media/musanmix.mp3', './media/musanmix.mp3'], lout)
self.assertTrue(filecmp.cmp(lout[0], lout[1]))
self.assertTrue(filecmp.cmp(lout[0], './media/musanmix-sm-gender.csv'))
def test_praat_export(self):
seg = Segmenter()
with tempfile.TemporaryDirectory() as tmpdirname:
lout = [os.path.join(tmpdirname, '1.TextGrid')]
ret = seg.batch_process(['./media/musanmix.mp3'], lout, output_format='textgrid')
self.assertTrue(filecmp.cmp(lout[0], './media/musanmix-smn-gender.TextGrid'))
import Pyro4
import sys
import os
import socket
from inaSpeechSegmenter import Segmenter
if __name__ == '__main__':
dname = os.path.dirname(os.path.realpath(__file__))
hostname = socket.gethostname()
uri = sys.argv[1]
jobserver = Pyro4.Proxy(uri)
ret = -1
outname = 'init'
# batch size set at 1024. Use lower values with small gpus
g = Segmenter(batch_size=1024)
while True:
lsrc, ldst = jobserver.get_njobs('%s %s' % (hostname, ret))
print(lsrc, ldst)
if len(lsrc) == 0:
print('job list finished')
break
ret = g.batch_process(lsrc, ldst, skipifexist=True, nbtry=3)
else:
input_files += glob.glob(e)
assert len(
input_files
) > 0, 'No existing media selected for analysis! Bad values provided to -i (%s)' % args.input
odir = args.output_directory.strip(" \t\n\r").rstrip('/')
assert os.access(odir, os.W_OK), 'Directory %s is not writable!' % odir
# Do processings
from inaSpeechSegmenter import Segmenter, seg2csv
# load neural network into memory, may last few seconds
detect_gender = bool(distutils.util.strtobool(args.detect_gender))
seg = Segmenter(vad_engine=args.vad_engine,
detect_gender=detect_gender,
ffmpeg=args.ffmpeg_binary,
energy_ratio=args.energy_ratio,
batch_size=args.batch_size)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
base = [os.path.splitext(os.path.basename(e))[0] for e in input_files]
output_files = [
os.path.join(odir, e + '.' + args.export_format) for e in base
]
seg.batch_process(input_files,
output_files,
verbose=True,
output_format=args.export_format)
args = parser.parse_args()
# Preprocess arguments and check their consistency
input_files = []
for e in args.input:
if e.startswith("http"):
input_files += [e]
else:
input_files += glob.glob(e)
assert len(
input_files
) > 0, 'No existing media selected for analysis! Bad values provided to -i (%s)' % args.input
odir = args.output_directory.strip(" \t\n\r").rstrip('/')
assert os.access(odir, os.W_OK), 'Directory %s is not writable!' % odir
# Do processings
from inaSpeechSegmenter import Segmenter, seg2csv
# load neural network into memory, may last few seconds
detect_gender = bool(distutils.util.strtobool(args.detect_gender))
seg = Segmenter(vad_engine=args.vad_engine,
detect_gender=detect_gender,
ffmpeg=args.ffmpeg_binary)
with warnings.catch_warnings():
warnings.simplefilter("ignore")
base = [os.path.splitext(os.path.basename(e))[0] for e in input_files]
output_files = [os.path.join(odir, e + '.csv') for e in base]
seg.batch_process(input_files, output_files, verbose=True)