mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-08 09:32:20 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix style issues. (#937)

Browse Source
This commit is contained in:
Fangjun Kuang 2023-03-08 22:56:04 +08:00 committed by GitHub
parent 07243d136a
commit f5de2e90c6
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
60 changed files with 185 additions and 552 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
egs/aidatatang_200zh/ASR/pruned_transducer_stateless2/decode.py
View File

@ -391,18 +391,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -412,9 +408,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/aishell/ASR/pruned_transducer_stateless2/decode.py
View File

@ -388,18 +388,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
# we compute CER for aishell dataset.
results_char = []
for res in results:
@ -413,9 +409,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/aishell/ASR/pruned_transducer_stateless3/decode.py
View File

@ -406,18 +406,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
# we compute CER for aishell dataset.
results_char = []
for res in results:
@ -431,9 +427,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tCER", file=f)
for key, val in test_set_wers:

12
egs/aishell/ASR/transducer_stateless/decode.py
View File

@ -325,17 +325,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
# we compute CER for aishell dataset.
results_char = []
for res in results:
@ -349,9 +345,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tCER", file=f)
for key, val in test_set_wers:

12
egs/aishell/ASR/transducer_stateless_modified-2/decode.py
View File

@ -370,18 +370,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
# we compute CER for aishell dataset.
results_char = []
for res in results:
@ -395,9 +391,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tCER", file=f)
for key, val in test_set_wers:

12
egs/aishell/ASR/transducer_stateless_modified/decode.py
View File

@ -374,18 +374,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
# we compute CER for aishell dataset.
results_char = []
for res in results:
@ -399,9 +395,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tCER", file=f)
for key, val in test_set_wers:

12
egs/aishell2/ASR/pruned_transducer_stateless5/decode.py
View File

@ -543,18 +543,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -564,9 +560,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/aishell4/ASR/pruned_transducer_stateless5/decode.py
View File

@ -406,18 +406,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -427,9 +423,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/alimeeting/ASR/pruned_transducer_stateless2/decode.py
View File

@ -391,18 +391,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -412,9 +408,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/alimeeting/ASR_v2/pruned_transducer_stateless7/decode.py
View File

@ -462,18 +462,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -483,9 +479,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

16
egs/ami/ASR/pruned_transducer_stateless7/decode.py
View File

@ -478,17 +478,13 @@ def save_results(
test_set_wers = dict()
test_set_cers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
wers_filename = (
params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt"
)
wers_filename = params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt"
with open(wers_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -499,9 +495,7 @@ def save_results(
results_char = []
for res in results:
results_char.append((res[0], list("".join(res[1])), list("".join(res[2]))))
cers_filename = (
params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt"
)
cers_filename = params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt"
with open(cers_filename, "w") as f:
cer = write_error_stats(
f, f"{test_set_name}-{key}", results_char, enable_log=True
@ -512,9 +506,7 @@ def save_results(
test_set_wers = {k: v for k, v in sorted(test_set_wers.items(), key=lambda x: x[1])}
test_set_cers = {k: v for k, v in sorted(test_set_cers.items(), key=lambda x: x[1])}
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER\tCER", file=f)
for key in test_set_wers:

12
egs/csj/ASR/pruned_transducer_stateless7_streaming/decode.py
View File

@ -599,9 +599,7 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
@ -609,9 +607,7 @@ def save_results(
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -621,9 +617,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/gigaspeech/ASR/pruned_transducer_stateless2/decode.py
View File

@ -399,9 +399,7 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = post_processing(results)
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
@ -409,9 +407,7 @@ def save_results(
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -421,9 +417,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

15
egs/librispeech/ASR/conformer_ctc3/decode.py
View File

@ -728,18 +728,14 @@ def save_results(
test_set_wers = dict()
test_set_delays = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts_and_timestamps(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer, mean_delay, var_delay = write_error_stats_with_timestamps(
f,
@ -754,9 +750,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:
@ -765,8 +759,7 @@ def save_results(
# sort according to the mean start symbol delay
test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0][0])
delays_info = (
params.res_dir
/ f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
)
with open(delays_info, "w") as f:
print("settings\t(start, end) symbol-delay (s) (start, end)", file=f)

12
egs/librispeech/ASR/conv_emformer_transducer_stateless/decode.py
View File

@ -432,18 +432,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -453,9 +449,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/conv_emformer_transducer_stateless/streaming_decode.py
View File

@ -750,17 +750,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=sorted(results))
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -770,9 +766,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/conv_emformer_transducer_stateless2/decode.py
View File

@ -432,18 +432,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -453,9 +449,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/conv_emformer_transducer_stateless2/streaming_decode.py
View File

@ -750,17 +750,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=sorted(results))
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -770,9 +766,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/lstm_transducer_stateless/decode.py
View File

@ -566,18 +566,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -587,9 +583,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/lstm_transducer_stateless/streaming_decode.py
View File

@ -742,17 +742,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=sorted(results))
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -762,9 +758,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/lstm_transducer_stateless2/decode.py
View File

@ -702,18 +702,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -723,9 +719,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

15
egs/librispeech/ASR/lstm_transducer_stateless3/decode.py
View File

@ -611,18 +611,14 @@ def save_results(
test_set_wers = dict()
test_set_delays = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts_and_timestamps(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer, mean_delay, var_delay = write_error_stats_with_timestamps(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -633,9 +629,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:
@ -643,8 +637,7 @@ def save_results(
test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0])
delays_info = (
params.res_dir
/ f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
)
with open(delays_info, "w") as f:
print("settings\tsymbol-delay", file=f)

12
egs/librispeech/ASR/lstm_transducer_stateless3/streaming_decode.py
View File

@ -742,17 +742,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=sorted(results))
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -762,9 +758,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned2_knowledge/decode.py
View File

@ -386,17 +386,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -406,9 +402,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_stateless_emformer_rnnt2/decode.py
View File

@ -420,18 +420,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -441,9 +437,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless/decode.py
View File

@ -585,18 +585,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -606,9 +602,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless/streaming_decode.py
View File

@ -423,9 +423,7 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
# sort results so we can easily compare the difference between two
# recognition results
results = sorted(results)
@ -434,9 +432,7 @@ def save_results(
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -446,9 +442,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless2/decode.py
View File

@ -609,18 +609,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -630,9 +626,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless2/streaming_decode.py
View File

@ -425,9 +425,7 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
# sort results so we can easily compare the difference between two
# recognition results
results = sorted(results)
@ -436,9 +434,7 @@ def save_results(
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -448,9 +444,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless3/decode.py
View File

@ -869,18 +869,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -890,9 +886,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless3/streaming_decode.py
View File

@ -426,18 +426,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -447,9 +443,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

15
egs/librispeech/ASR/pruned_transducer_stateless4/decode.py
View File

@ -655,18 +655,14 @@ def save_results(
test_set_wers = dict()
test_set_delays = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts_and_timestamps(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer, mean_delay, var_delay = write_error_stats_with_timestamps(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -677,9 +673,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:
@ -687,8 +681,7 @@ def save_results(
test_set_delays = sorted(test_set_delays.items(), key=lambda x: x[1][0])
delays_info = (
params.res_dir
/ f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
params.res_dir / f"symbol-delay-summary-{test_set_name}-{params.suffix}.txt"
)
with open(delays_info, "w") as f:
print("settings\tsymbol-delay", file=f)

12
egs/librispeech/ASR/pruned_transducer_stateless4/streaming_decode.py
View File

@ -442,18 +442,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -463,9 +459,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless5/decode.py
View File

@ -735,18 +735,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -756,9 +752,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless5/streaming_decode.py
View File

@ -442,18 +442,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -463,9 +459,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless6/decode.py
View File

@ -416,18 +416,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -437,9 +433,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7/decode.py
View File

@ -722,18 +722,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -743,9 +739,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_ctc/ctc_decode.py
View File

@ -541,18 +541,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(f, f"{test_set_name}-{key}", results)
test_set_wers[key] = wer
@ -560,9 +556,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_ctc/decode.py
View File

@ -593,18 +593,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -614,9 +610,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/ctc_decode.py
View File

@ -532,18 +532,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(f, f"{test_set_name}-{key}", results)
test_set_wers[key] = wer
@ -551,9 +547,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_ctc_bs/decode.py
View File

@ -593,18 +593,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -614,9 +610,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_streaming/decode.py
View File

@ -568,18 +568,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -589,9 +585,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless7_streaming/streaming_decode.py
View File

@ -409,18 +409,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -430,9 +426,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/pruned_transducer_stateless8/decode.py
View File

@ -594,18 +594,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -615,9 +611,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/transducer/decode.py
View File

@ -325,18 +325,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -346,9 +342,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/transducer_lstm/decode.py
View File

@ -322,18 +322,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -343,9 +339,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/transducer_stateless/decode.py
View File

@ -379,18 +379,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -400,9 +396,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/transducer_stateless2/decode.py
View File

@ -379,18 +379,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -400,9 +396,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/transducer_stateless_multi_datasets/decode.py
View File

@ -380,18 +380,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -401,9 +397,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/librispeech/ASR/zipformer_mmi/decode.py
View File

@ -471,18 +471,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(f, f"{test_set_name}-{key}", results)
test_set_wers[key] = wer
@ -490,9 +486,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/mgb2/ASR/pruned_transducer_stateless5/decode.py
View File

@ -410,17 +410,13 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -430,9 +426,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

16
egs/spgispeech/ASR/pruned_transducer_stateless2/decode.py
View File

@ -391,18 +391,14 @@ def save_results(
test_set_wers = dict()
test_set_cers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
wers_filename = (
params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt"
)
wers_filename = params.res_dir / f"wers-{test_set_name}-{params.suffix}.txt"
with open(wers_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -413,9 +409,7 @@ def save_results(
results_char = []
for res in results:
results_char.append((res[0], list("".join(res[1])), list("".join(res[2]))))
cers_filename = (
params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt"
)
cers_filename = params.res_dir / f"cers-{test_set_name}-{params.suffix}.txt"
with open(cers_filename, "w") as f:
cer = write_error_stats(
f, f"{test_set_name}-{key}", results_char, enable_log=True
@ -426,9 +420,7 @@ def save_results(
test_set_wers = {k: v for k, v in sorted(test_set_wers.items(), key=lambda x: x[1])}
test_set_cers = {k: v for k, v in sorted(test_set_cers.items(), key=lambda x: x[1])}
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER\tCER", file=f)
for key in test_set_wers:

12
egs/tal_csasr/ASR/pruned_transducer_stateless5/decode.py
View File

@ -509,18 +509,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -530,9 +526,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/tedlium3/ASR/pruned_transducer_stateless/decode.py
View File

@ -379,18 +379,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -400,9 +396,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/tedlium3/ASR/transducer_stateless/decode.py
View File

@ -354,18 +354,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -375,9 +371,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/wenetspeech/ASR/pruned_transducer_stateless2/decode.py
View File

@ -516,18 +516,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -537,9 +533,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/wenetspeech/ASR/pruned_transducer_stateless5/decode.py
View File

@ -489,18 +489,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -510,9 +506,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/wenetspeech/ASR/pruned_transducer_stateless5/streaming_decode.py
View File

@ -466,9 +466,7 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
# sort results so we can easily compare the difference between two
# recognition results
results = sorted(results)
@ -477,9 +475,7 @@ def save_results(
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -489,9 +485,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/xbmu_amdo31/ASR/pruned_transducer_stateless5/decode.py
View File

@ -701,18 +701,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -722,9 +718,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers:

12
egs/xbmu_amdo31/ASR/pruned_transducer_stateless7/decode.py
View File

@ -593,18 +593,14 @@ def save_results(
):
test_set_wers = dict()
for key, results in results_dict.items():
recog_path = (
params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
)
recog_path = params.res_dir / f"recogs-{test_set_name}-{params.suffix}.txt"
results = sorted(results)
store_transcripts(filename=recog_path, texts=results)
logging.info(f"The transcripts are stored in {recog_path}")
# The following prints out WERs, per-word error statistics and aligned
# ref/hyp pairs.
errs_filename = (
params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
)
errs_filename = params.res_dir / f"errs-{test_set_name}-{params.suffix}.txt"
with open(errs_filename, "w") as f:
wer = write_error_stats(
f, f"{test_set_name}-{key}", results, enable_log=True
@ -614,9 +610,7 @@ def save_results(
logging.info("Wrote detailed error stats to {}".format(errs_filename))
test_set_wers = sorted(test_set_wers.items(), key=lambda x: x[1])
errs_info = (
params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
)
errs_info = params.res_dir / f"wer-summary-{test_set_name}-{params.suffix}.txt"
with open(errs_info, "w") as f:
print("settings\tWER", file=f)
for key, val in test_set_wers: