From 9903ed68237a2a353f12eb136d845d9637d83e6d Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期三, 24 五月 2023 10:58:26 +0800
Subject: [PATCH] Merge branch 'main' of github.com:alibaba-damo-academy/FunASR merge
---
docs/m2met2/_build/html/_images/qrcode.png | 0
docs/m2met2_cn/_build/doctrees/简介.doctree | 0
docs/m2met2_cn/_build/doctrees/联系方式.doctree | 0
egs/alimeeting/sa-asr/asr_local.sh | 8 ++--
funasr/bin/asr_train.py | 7 ++-
docs/m2met2_cn/_build/html/.buildinfo | 2
docs/m2met2/_build/doctrees/Track_setting_and_evaluation.doctree | 0
docs/m2met2_cn/_build/doctrees/规则.doctree | 0
docs/m2met2/_build/doctrees/Dataset.doctree | 0
docs/m2met2_cn/_build/doctrees/index.doctree | 0
docs/m2met2_cn/images/qrcode.png | 0
docs/m2met2/_build/doctrees/Baseline.doctree | 0
docs/m2met2/_build/doctrees/Organizers.doctree | 0
docs/m2met2_cn/_build/doctrees/environment.pickle | 0
docs/m2met2/_build/doctrees/Rules.doctree | 0
docs/m2met2_cn/_build/doctrees/基线.doctree | 0
docs/m2met2_cn/_build/doctrees/组委会.doctree | 0
egs/alimeeting/sa-asr/conf/train_asr_conformer.yaml | 1
docs/m2met2/_build/html/.buildinfo | 2
docs/m2met2/_build/doctrees/environment.pickle | 0
docs/m2met2/_build/doctrees/index.doctree | 0
docs/m2met2_cn/_build/doctrees/赛道设置与评估.doctree | 0
funasr/tasks/asr.py | 9 +---
funasr/tasks/abs_task.py | 23 ++---------
egs/alimeeting/sa-asr/conf/train_sa_asr_conformer.yaml | 1
docs/m2met2/_build/doctrees/Contact.doctree | 0
docs/m2met2/images/qrcode.png | 0
docs/m2met2_cn/_build/html/_images/qrcode.png | 0
funasr/bin/asr_infer.py | 22 ++++++++---
docs/m2met2_cn/_build/doctrees/数据集.doctree | 0
docs/m2met2/_build/doctrees/Introduction.doctree | 0
31 files changed, 34 insertions(+), 41 deletions(-)
diff --git a/docs/m2met2/_build/doctrees/Baseline.doctree b/docs/m2met2/_build/doctrees/Baseline.doctree
index c9c4075..e9db7d5 100644
--- a/docs/m2met2/_build/doctrees/Baseline.doctree
+++ b/docs/m2met2/_build/doctrees/Baseline.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Contact.doctree b/docs/m2met2/_build/doctrees/Contact.doctree
index 0508819..e272be7 100644
--- a/docs/m2met2/_build/doctrees/Contact.doctree
+++ b/docs/m2met2/_build/doctrees/Contact.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Dataset.doctree b/docs/m2met2/_build/doctrees/Dataset.doctree
index 47c8bb4..7030064 100644
--- a/docs/m2met2/_build/doctrees/Dataset.doctree
+++ b/docs/m2met2/_build/doctrees/Dataset.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Introduction.doctree b/docs/m2met2/_build/doctrees/Introduction.doctree
index 6ffceef..bbf9a37 100644
--- a/docs/m2met2/_build/doctrees/Introduction.doctree
+++ b/docs/m2met2/_build/doctrees/Introduction.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Organizers.doctree b/docs/m2met2/_build/doctrees/Organizers.doctree
index 7ecfbdf..87a23b8 100644
--- a/docs/m2met2/_build/doctrees/Organizers.doctree
+++ b/docs/m2met2/_build/doctrees/Organizers.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Rules.doctree b/docs/m2met2/_build/doctrees/Rules.doctree
index 5b7655f..ac881da 100644
--- a/docs/m2met2/_build/doctrees/Rules.doctree
+++ b/docs/m2met2/_build/doctrees/Rules.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/Track_setting_and_evaluation.doctree b/docs/m2met2/_build/doctrees/Track_setting_and_evaluation.doctree
index 8770fba..4fdd68a 100644
--- a/docs/m2met2/_build/doctrees/Track_setting_and_evaluation.doctree
+++ b/docs/m2met2/_build/doctrees/Track_setting_and_evaluation.doctree
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/environment.pickle b/docs/m2met2/_build/doctrees/environment.pickle
index 6fa2888..3002d02 100644
--- a/docs/m2met2/_build/doctrees/environment.pickle
+++ b/docs/m2met2/_build/doctrees/environment.pickle
Binary files differ
diff --git a/docs/m2met2/_build/doctrees/index.doctree b/docs/m2met2/_build/doctrees/index.doctree
index f40d4fe..9469f3c 100644
--- a/docs/m2met2/_build/doctrees/index.doctree
+++ b/docs/m2met2/_build/doctrees/index.doctree
Binary files differ
diff --git a/docs/m2met2/_build/html/.buildinfo b/docs/m2met2/_build/html/.buildinfo
index 97d32c4..4ed53f5 100644
--- a/docs/m2met2/_build/html/.buildinfo
+++ b/docs/m2met2/_build/html/.buildinfo
@@ -1,4 +1,4 @@
# Sphinx build info version 1
# This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: a62852d90c3e533904d811bbf85f977d
+config: 160d25833895e2f6c62a4c315cacc3b9
tags: 645f666f9bcd5a90fca523b33c5a78b7
diff --git a/docs/m2met2/_build/html/_images/qrcode.png b/docs/m2met2/_build/html/_images/qrcode.png
index fc4c349..d18d44a 100644
--- a/docs/m2met2/_build/html/_images/qrcode.png
+++ b/docs/m2met2/_build/html/_images/qrcode.png
Binary files differ
diff --git a/docs/m2met2/images/qrcode.png b/docs/m2met2/images/qrcode.png
index fc4c349..d18d44a 100644
--- a/docs/m2met2/images/qrcode.png
+++ b/docs/m2met2/images/qrcode.png
Binary files differ
diff --git a/docs/m2met2_cn/_build/doctrees/environment.pickle b/docs/m2met2_cn/_build/doctrees/environment.pickle
index 377374e..a65d613 100644
--- a/docs/m2met2_cn/_build/doctrees/environment.pickle
+++ b/docs/m2met2_cn/_build/doctrees/environment.pickle
Binary files differ
diff --git a/docs/m2met2_cn/_build/doctrees/index.doctree b/docs/m2met2_cn/_build/doctrees/index.doctree
index 1677b8b..43e1b34 100644
--- a/docs/m2met2_cn/_build/doctrees/index.doctree
+++ b/docs/m2met2_cn/_build/doctrees/index.doctree
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\345\237\272\347\272\277.doctree" "b/docs/m2met2_cn/_build/doctrees/\345\237\272\347\272\277.doctree"
index e31d3dd..aa459c1 100644
--- "a/docs/m2met2_cn/_build/doctrees/\345\237\272\347\272\277.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\345\237\272\347\272\277.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\346\225\260\346\215\256\351\233\206.doctree" "b/docs/m2met2_cn/_build/doctrees/\346\225\260\346\215\256\351\233\206.doctree"
index 5950ee3..2bc6bd7 100644
--- "a/docs/m2met2_cn/_build/doctrees/\346\225\260\346\215\256\351\233\206.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\346\225\260\346\215\256\351\233\206.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\347\256\200\344\273\213.doctree" "b/docs/m2met2_cn/_build/doctrees/\347\256\200\344\273\213.doctree"
index abe249d..6d6aab7 100644
--- "a/docs/m2met2_cn/_build/doctrees/\347\256\200\344\273\213.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\347\256\200\344\273\213.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\347\273\204\345\247\224\344\274\232.doctree" "b/docs/m2met2_cn/_build/doctrees/\347\273\204\345\247\224\344\274\232.doctree"
index 551248c..dbec226 100644
--- "a/docs/m2met2_cn/_build/doctrees/\347\273\204\345\247\224\344\274\232.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\347\273\204\345\247\224\344\274\232.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\350\201\224\347\263\273\346\226\271\345\274\217.doctree" "b/docs/m2met2_cn/_build/doctrees/\350\201\224\347\263\273\346\226\271\345\274\217.doctree"
index 6b8208a..dfdd79a 100644
--- "a/docs/m2met2_cn/_build/doctrees/\350\201\224\347\263\273\346\226\271\345\274\217.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\350\201\224\347\263\273\346\226\271\345\274\217.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\350\247\204\345\210\231.doctree" "b/docs/m2met2_cn/_build/doctrees/\350\247\204\345\210\231.doctree"
index 1cf759c..b74e39e 100644
--- "a/docs/m2met2_cn/_build/doctrees/\350\247\204\345\210\231.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\350\247\204\345\210\231.doctree"
Binary files differ
diff --git "a/docs/m2met2_cn/_build/doctrees/\350\265\233\351\201\223\350\256\276\347\275\256\344\270\216\350\257\204\344\274\260.doctree" "b/docs/m2met2_cn/_build/doctrees/\350\265\233\351\201\223\350\256\276\347\275\256\344\270\216\350\257\204\344\274\260.doctree"
index c6be4ae..35e0a06 100644
--- "a/docs/m2met2_cn/_build/doctrees/\350\265\233\351\201\223\350\256\276\347\275\256\344\270\216\350\257\204\344\274\260.doctree"
+++ "b/docs/m2met2_cn/_build/doctrees/\350\265\233\351\201\223\350\256\276\347\275\256\344\270\216\350\257\204\344\274\260.doctree"
Binary files differ
diff --git a/docs/m2met2_cn/_build/html/.buildinfo b/docs/m2met2_cn/_build/html/.buildinfo
index 35632ee..43b8395 100644
--- a/docs/m2met2_cn/_build/html/.buildinfo
+++ b/docs/m2met2_cn/_build/html/.buildinfo
@@ -1,4 +1,4 @@
# Sphinx build info version 1
# This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: 06d9c1d4093817b45b9d4df7ab350eaf
+config: a4d4595bd4f85adbedc556dc23e6150a
tags: 645f666f9bcd5a90fca523b33c5a78b7
diff --git a/docs/m2met2_cn/_build/html/_images/qrcode.png b/docs/m2met2_cn/_build/html/_images/qrcode.png
index fc4c349..d18d44a 100644
--- a/docs/m2met2_cn/_build/html/_images/qrcode.png
+++ b/docs/m2met2_cn/_build/html/_images/qrcode.png
Binary files differ
diff --git a/docs/m2met2_cn/images/qrcode.png b/docs/m2met2_cn/images/qrcode.png
index fc4c349..d18d44a 100644
--- a/docs/m2met2_cn/images/qrcode.png
+++ b/docs/m2met2_cn/images/qrcode.png
Binary files differ
diff --git a/egs/alimeeting/sa-asr/asr_local.sh b/egs/alimeeting/sa-asr/asr_local.sh
index 05599b7..30401b9 100755
--- a/egs/alimeeting/sa-asr/asr_local.sh
+++ b/egs/alimeeting/sa-asr/asr_local.sh
@@ -1153,10 +1153,10 @@
mkdir -p ${sa_asr_exp}/log
INIT_FILE=${sa_asr_exp}/ddp_init
- if [ ! -f "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pb" ]; then
+ if [ ! -f "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pth" ]; then
# download xvector extractor model file
python local/download_xvector_model.py exp
- log "Successfully download the pretrained xvector extractor to exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pb"
+ log "Successfully download the pretrained xvector extractor to exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pth"
fi
if [ -f $INIT_FILE ];then
@@ -1195,8 +1195,8 @@
--init_param "${asr_exp}/valid.acc.ave.pb:decoder.decoders.3:decoder.decoder4.2" \
--init_param "${asr_exp}/valid.acc.ave.pb:decoder.decoders.4:decoder.decoder4.3" \
--init_param "${asr_exp}/valid.acc.ave.pb:decoder.decoders.5:decoder.decoder4.4" \
- --init_param "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pb:encoder:spk_encoder" \
- --init_param "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pb:decoder:spk_encoder:decoder.output_dense" \
+ --init_param "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pth:encoder:spk_encoder" \
+ --init_param "exp/damo/speech_xvector_sv-zh-cn-cnceleb-16k-spk3465-pytorch/sv.pth:decoder:spk_encoder:decoder.output_dense" \
--valid_data_path_and_name_and_type "${_asr_valid_dir}/${_scp},speech,${_type}" \
--valid_data_path_and_name_and_type "${_asr_valid_dir}/text,text,text" \
--valid_data_path_and_name_and_type "${_asr_valid_dir}/oracle_profile_nopadding.scp,profile,npy" \
diff --git a/egs/alimeeting/sa-asr/conf/train_asr_conformer.yaml b/egs/alimeeting/sa-asr/conf/train_asr_conformer.yaml
index 7865763..a8c9968 100644
--- a/egs/alimeeting/sa-asr/conf/train_asr_conformer.yaml
+++ b/egs/alimeeting/sa-asr/conf/train_asr_conformer.yaml
@@ -4,6 +4,7 @@
n_fft: 400
win_length: 400
hop_length: 160
+ use_channel: 0
# encoder related
encoder: conformer
diff --git a/egs/alimeeting/sa-asr/conf/train_sa_asr_conformer.yaml b/egs/alimeeting/sa-asr/conf/train_sa_asr_conformer.yaml
index aa48b2d..612fd23 100644
--- a/egs/alimeeting/sa-asr/conf/train_sa_asr_conformer.yaml
+++ b/egs/alimeeting/sa-asr/conf/train_sa_asr_conformer.yaml
@@ -4,6 +4,7 @@
n_fft: 400
win_length: 400
hop_length: 160
+ use_channel: 0
# encoder related
asr_encoder: conformer
diff --git a/funasr/bin/asr_infer.py b/funasr/bin/asr_infer.py
index 9da7ef7..760fd07 100644
--- a/funasr/bin/asr_infer.py
+++ b/funasr/bin/asr_infer.py
@@ -1510,8 +1510,13 @@
if isinstance(speech, np.ndarray):
speech = torch.tensor(speech)
- feats = speech.unsqueeze(0).to(getattr(torch, self.dtype))
- feats_lengths = feats.new_full([1], dtype=torch.long, fill_value=feats.size(1))
+ if self.frontend is not None:
+ speech = torch.unsqueeze(speech, axis=0)
+ speech_lengths = speech.new_full([1], dtype=torch.long, fill_value=speech.size(1))
+ feats, feats_lengths = self.frontend(speech, speech_lengths)
+ else:
+ feats = speech.unsqueeze(0).to(getattr(torch, self.dtype))
+ feats_lengths = feats.new_full([1], dtype=torch.long, fill_value=feats.size(1))
if self.asr_model.normalize is not None:
feats, feats_lengths = self.asr_model.normalize(feats, feats_lengths)
@@ -1536,14 +1541,19 @@
if isinstance(speech, np.ndarray):
speech = torch.tensor(speech)
-
- feats = speech.unsqueeze(0).to(getattr(torch, self.dtype))
- feats_lengths = feats.new_full([1], dtype=torch.long, fill_value=feats.size(1))
+
+ if self.frontend is not None:
+ speech = torch.unsqueeze(speech, axis=0)
+ speech_lengths = speech.new_full([1], dtype=torch.long, fill_value=speech.size(1))
+ feats, feats_lengths = self.frontend(speech, speech_lengths)
+ else:
+ feats = speech.unsqueeze(0).to(getattr(torch, self.dtype))
+ feats_lengths = feats.new_full([1], dtype=torch.long, fill_value=feats.size(1))
feats = to_device(feats, device=self.device)
feats_lengths = to_device(feats_lengths, device=self.device)
- enc_out, _ = self.asr_model.encoder(feats, feats_lengths)
+ enc_out, _, _ = self.asr_model.encoder(feats, feats_lengths)
nbest_hyps = self.beam_search(enc_out[0])
diff --git a/funasr/bin/asr_train.py b/funasr/bin/asr_train.py
index fd973a4..8161e7b 100755
--- a/funasr/bin/asr_train.py
+++ b/funasr/bin/asr_train.py
@@ -46,7 +46,8 @@
args = parse_args()
# setup local gpu_id
- os.environ['CUDA_VISIBLE_DEVICES'] = str(args.gpu_id)
+ if args.ngpu > 0:
+ os.environ['CUDA_VISIBLE_DEVICES'] = str(args.gpu_id)
# DDP settings
if args.ngpu > 1:
@@ -57,9 +58,9 @@
# re-compute batch size: when dataset type is small
if args.dataset_type == "small":
- if args.batch_size is not None:
+ if args.batch_size is not None and args.ngpu > 0:
args.batch_size = args.batch_size * args.ngpu
- if args.batch_bins is not None:
+ if args.batch_bins is not None and args.ngpu > 0:
args.batch_bins = args.batch_bins * args.ngpu
main(args=args)
diff --git a/funasr/tasks/abs_task.py b/funasr/tasks/abs_task.py
index fd4e190..0fb77a9 100644
--- a/funasr/tasks/abs_task.py
+++ b/funasr/tasks/abs_task.py
@@ -1376,25 +1376,10 @@
# 7. Build iterator factories
if args.dataset_type == "large":
- from funasr.datasets.large_datasets.build_dataloader import ArkDataLoader
- train_iter_factory = ArkDataLoader(args.train_data_file, args.token_list, args.dataset_conf,
- frontend_conf=args.frontend_conf if hasattr(args,
- "frontend_conf") else None,
- seg_dict_file=args.seg_dict_file if hasattr(args,
- "seg_dict_file") else None,
- punc_dict_file=args.punc_list if hasattr(args,
- "punc_list") else None,
- bpemodel_file=args.bpemodel if hasattr(args, "bpemodel") else None,
- mode="train")
- valid_iter_factory = ArkDataLoader(args.valid_data_file, args.token_list, args.dataset_conf,
- frontend_conf=args.frontend_conf if hasattr(args,
- "frontend_conf") else None,
- seg_dict_file=args.seg_dict_file if hasattr(args,
- "seg_dict_file") else None,
- punc_dict_file=args.punc_list if hasattr(args,
- "punc_list") else None,
- bpemodel_file=args.bpemodel if hasattr(args, "bpemodel") else None,
- mode="eval")
+ from funasr.datasets.large_datasets.build_dataloader import LargeDataLoader
+ train_iter_factory = LargeDataLoader(args, mode="train")
+ valid_iter_factory = LargeDataLoader(args, mode="eval")
+
elif args.dataset_type == "small":
train_iter_factory = cls.build_iter_factory(
args=args,
diff --git a/funasr/tasks/asr.py b/funasr/tasks/asr.py
index 8e4f9cc..8244856 100644
--- a/funasr/tasks/asr.py
+++ b/funasr/tasks/asr.py
@@ -363,12 +363,6 @@
default=get_default_kwargs(CTC),
help="The keyword arguments for CTC class.",
)
- group.add_argument(
- "--joint_network_conf",
- action=NestedDictAction,
- default=None,
- help="The keyword arguments for joint network class.",
- )
group = parser.add_argument_group(description="Preprocess related")
group.add_argument(
@@ -1379,6 +1373,7 @@
num_optimizers: int = 1
class_choices_list = [
+ model_choices,
frontend_choices,
specaug_choices,
normalize_choices,
@@ -1476,7 +1471,7 @@
try:
model_class = model_choices.get_class(args.model)
except AttributeError:
- model_class = model_choices.get_class("asr")
+ model_class = model_choices.get_class("rnnt_unified")
model = model_class(
vocab_size=vocab_size,
--
Gitblit v1.9.1