help="# of process for Multiprocessing",
default=8,
type=int)
args = parser.parse_args()
assert os.path.isdir(str(
args.wsj0)), "WSJ0 directory not found - '{d}'".format(d=args.wsj0)
assert os.path.isdir(str(
args.wsj1)), "WSJ1 directory not found - '{d}'".format(d=args.wsj1)
assert args.wsj0 != args.wsj1, "WSJ0 and WSJ1 directories can't be the same"
assert os.path.exists(
args.sph2pipe), "sph2pipe not found '{d}'".format(d=args.sph2pipe)
transcripts = {}
utils.find_transcripts(args.wsj0, transcripts)
utils.find_transcripts(args.wsj1, transcripts)
sets = {}
sets["si84"] = utils.ndx2idlist(
args.wsj0,
"11-13.1/wsj0/doc/indices/train/tr_s_wv1.ndx",
transcripts,
lambda line: None if "11_2_1:wsj0/si_tr_s/401" in line else line,
)
assert len(sets["si84"]) == 7138
sets["si284"] = utils.ndx2idlist(
args.wsj0,
"11-13.1/wsj0/doc/indices/train/tr_s_wv1.ndx",
transcripts,
default=8,
type=int)
args = parser.parse_args()
wsj1_sep = "-" if args.wsj1_type == "LDC94S13A" else "_"
assert os.path.isdir(str(
args.wsj0)), "WSJ0 directory is not found - '{d}'".format(d=args.wsj0)
assert os.path.isdir(str(
args.wsj1)), "WSJ1 directory is not found - '{d}'".format(d=args.wsj1)
assert args.wsj0 != args.wsj1, "WSJ0 and WSJ1 directories can't be the same"
assert os.path.exists(
args.sph2pipe), "sph2pipe not found '{d}'".format(d=args.sph2pipe)
# Prepare audio data
transcripts = find_transcripts([args.wsj0, args.wsj1])
subsets = dict()
subsets["si84"] = ndx_to_samples(
args.wsj0,
"11-13.1/wsj0/doc/indices/train/tr_s_wv1.ndx",
transcripts,
lambda line: None if "11_2_1:wsj0/si_tr_s/401" in line else line,
)
assert len(
subsets["si84"]) == 7138, "Incorrect number of samples in si84 part:"
" should be 7138, but fould #{}.".format(len(subsets["si84"]))
subsets["si284"] = ndx_to_samples(
args.wsj0,
"11-13.1/wsj0/doc/indices/train/tr_s_wv1.ndx",