Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Add few shot subset for mSuperb multilingual setting #4923

Merged
merged 3 commits into from
Feb 8, 2023
Merged

Add few shot subset for mSuperb multilingual setting #4923

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
3a798c3
Add few shot subset for mSuperb multilingual setting
guapaQAQ Feb 7, 2023
96350bc
fix black
guapaQAQ Feb 8, 2023
7950964
Merge branch 'master' into mSUPERB
guapaQAQ Feb 8, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
264 changes: 203 additions & 61 deletions egs2/msuperb/asr1/local/data_prep.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,11 +38,154 @@
"ceb",
"luo",
"kea",
"sum",
"sun",
"tso",
"tos",
]

FEW_SHOT_SELECTED_DATA = {
"lit": [
"cv_lit_000026",
"cv_lit_000105",
"fleurs_lit_000097",
"mls_lit_000040",
"voxpopuli_lit_001707",
],
"dan": [
"NST_dan_000007",
"NST_dan_000072",
"cv_dan_000158",
"fleurs_dan_000058",
"fleurs_dan_000071",
],
"tur": [
"cv_tur_000095",
"cv_tur_000101",
"fleurs_tur_000060",
"fleurs_tur_000069",
"fleurs_tur_000095",
],
"srp": [
"cv_srp_000078",
"cv_srp_000145",
"fleurs_srp_000068",
"fleurs_srp_000093",
"fleurs_srp_000105",
],
"vie": [
"cv_vie_000105",
"cv_vie_000128",
"fleurs_vie_000067",
"fleurs_vie_000068",
"fleurs_vie_000077",
],
"kaz": [
"cv_kaz_000080",
"cv_kaz_000097",
"cv_kaz_000111",
"fleurs_kaz_000036",
"fleurs_kaz_000066",
],
"zul": [
"fleurs_zul_000049",
"fleurs_zul_000057",
"nchlt_zul_000027",
"nchlt_zul_000090",
"nchlt_zul_000104",
],
"tsn": [
"googlei18n-tts_tsn_000026",
"googlei18n-tts_tsn_000044",
"googlei18n-tts_tsn_000108",
"nchlt_tsn_000001",
"nchlt_tsn_000032",
],
"epo": [
"cv_epo_000006",
"cv_epo_000039",
"cv_epo_000063",
"cv_epo_000066",
"cv_epo_000076",
],
"frr": [
"cv_frr_000023",
"cv_frr_000086",
"cv_frr_000095",
"cv_frr_000102",
"cv_frr_000104",
],
"tok": [
"cv_tok_000004",
"cv_tok_000011",
"cv_tok_000030",
"cv_tok_000084",
"cv_tok_000101",
],
"umb": [
"fleurs_umb_000028",
"fleurs_umb_000029",
"fleurs_umb_000033",
"fleurs_umb_000040",
"fleurs_umb_000047",
],
"bos": [
"fleurs_bos_000067",
"fleurs_bos_000078",
"fleurs_bos_000080",
"fleurs_bos_000088",
"fleurs_bos_000090",
],
"ful": [
"fleurs_ful_000055",
"fleurs_ful_000059",
"fleurs_ful_000067",
"fleurs_ful_000076",
"fleurs_ful_000081",
],
"ceb": [
"fleurs_ceb_000054",
"fleurs_ceb_000064",
"fleurs_ceb_000069",
"fleurs_ceb_000071",
"fleurs_ceb_000080",
],
"luo": [
"fleurs_luo_000056",
"fleurs_luo_000062",
"fleurs_luo_000067",
"fleurs_luo_000073",
"fleurs_luo_000077",
],
"kea": [
"fleurs_kea_000052",
"fleurs_kea_000070",
"fleurs_kea_000078",
"fleurs_kea_000083",
"fleurs_kea_000085",
],
"sun": [
"googlei18n-asr_sun_000001",
"googlei18n-asr_sun_000007",
"googlei18n-asr_sun_000041",
"googlei18n-asr_sun_000099",
"googlei18n-asr_sun_000106",
],
"tso": [
"nchlt_tso_000035",
"nchlt_tso_000040",
"nchlt_tso_000089",
"nchlt_tso_000104",
"nchlt_tso_000125",
],
"tos": [
"mexico-el_tos_000006",
"mexico-el_tos_000122",
"mexico-el_tos_000152",
"mexico-el_tos_000496",
"mexico-el_tos_000563",
],
}

if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--train_set", type=str, default="train_10min")
Expand Down Expand Up @@ -95,69 +238,68 @@
langs_info[lang] = []
langs_info[lang].append(dataset)

if not reserve_flag:
# process train
train_transcript = open(
os.path.join(
args.source,
dataset,
lang,
"transcript_{}_train.txt".format(args.duration),
),
"r",
encoding="utf-8",
)
for line in train_transcript.readlines():
line = line.strip().split(maxsplit=2)
utt_id, _, text = line
train_wavscp.write(
"{} sox {} -c 1 -t wavpcm -|\n".format(
utt_id,
os.path.join(
args.source,
dataset,
lang,
"wav",
"{}.wav".format(utt_id),
),
)
# process train
train_transcript = open(
os.path.join(
args.source,
dataset,
lang,
"transcript_{}_train.txt".format(args.duration),
),
"r",
encoding="utf-8",
)
for line in train_transcript.readlines():
line = line.strip().split(maxsplit=2)
utt_id, _, text = line
if reserve_flag and utt_id not in FEW_SHOT_SELECTED_DATA[lang]:
continue
train_wavscp.write(
"{} sox {} -c 1 -t wavpcm -|\n".format(
utt_id,
os.path.join(
args.source,
dataset,
lang,
"wav",
"{}.wav".format(utt_id),
),
)
if args.lid:
train_text.write("{} [{}] {}\n".format(utt_id, lang, text))
else:
train_text.write("{} {}\n".format(utt_id, text))
train_utt2spk.write("{} {}\n".format(utt_id, utt_id))
train_transcript.close()

# process dev
dev_transcript = open(
os.path.join(
args.source, dataset, lang, "transcript_10min_dev.txt"
),
"r",
encoding="utf-8",
)
for line in dev_transcript.readlines():
line = line.strip().split(maxsplit=2)
utt_id, _, text = line
dev_wavscp.write(
"{} sox {} -c 1 -t wavpcm -|\n".format(
utt_id,
os.path.join(
args.source,
dataset,
lang,
"wav",
"{}.wav".format(utt_id),
),
)
if args.lid:
train_text.write("{} [{}] {}\n".format(utt_id, lang, text))
else:
train_text.write("{} {}\n".format(utt_id, text))
train_utt2spk.write("{} {}\n".format(utt_id, utt_id))
train_transcript.close()

# process dev
dev_transcript = open(
os.path.join(args.source, dataset, lang, "transcript_10min_dev.txt"),
"r",
encoding="utf-8",
)
for line in dev_transcript.readlines():
line = line.strip().split(maxsplit=2)
utt_id, _, text = line
dev_wavscp.write(
"{} sox {} -c 1 -t wavpcm -|\n".format(
utt_id,
os.path.join(
args.source,
dataset,
lang,
"wav",
"{}.wav".format(utt_id),
),
)
if args.lid:
dev_text.write("{} [{}] {}\n".format(utt_id, lang, text))
else:
dev_text.write("{} {}\n".format(utt_id, text))
dev_utt2spk.write("{} {}\n".format(utt_id, utt_id))
dev_transcript.close()
)
if args.lid:
dev_text.write("{} [{}] {}\n".format(utt_id, lang, text))
else:
dev_text.write("{} {}\n".format(utt_id, text))
dev_utt2spk.write("{} {}\n".format(utt_id, utt_id))
dev_transcript.close()

# process test
test_transcript = open(
Expand Down