Skip to content
体验新版
项目
组织
正在加载...
登录
切换导航
打开侧边栏
PaddlePaddle
DeepSpeech
提交
97965f4c
D
DeepSpeech
项目概览
PaddlePaddle
/
DeepSpeech
大约 2 年 前同步成功
通知
210
Star
8425
Fork
1598
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
245
列表
看板
标记
里程碑
合并请求
3
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
D
DeepSpeech
项目概览
项目概览
详情
发布
仓库
仓库
文件
提交
分支
标签
贡献者
分支图
比较
Issue
245
Issue
245
列表
看板
标记
里程碑
合并请求
3
合并请求
3
Pages
分析
分析
仓库分析
DevOps
Wiki
0
Wiki
成员
成员
收起侧边栏
关闭侧边栏
动态
分支图
创建新Issue
提交
Issue看板
提交
97965f4c
编写于
7月 15, 2022
作者:
小湉湉
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
fix mlm_prob, test=tts
上级
c1395e3a
变更
7
隐藏空白更改
内联
并排
Showing
7 changed file
with
12 addition
and
21 deletion
+12
-21
examples/aishell3/ernie_sat/local/train.sh
examples/aishell3/ernie_sat/local/train.sh
+1
-1
examples/aishell3_vctk/ernie_sat/local/train.sh
examples/aishell3_vctk/ernie_sat/local/train.sh
+1
-1
examples/vctk/ernie_sat/conf/default.yaml
examples/vctk/ernie_sat/conf/default.yaml
+2
-2
examples/vctk/ernie_sat/local/train.sh
examples/vctk/ernie_sat/local/train.sh
+1
-1
paddlespeech/t2s/datasets/am_batch_fn.py
paddlespeech/t2s/datasets/am_batch_fn.py
+5
-12
paddlespeech/t2s/exps/ernie_sat/synthesize.py
paddlespeech/t2s/exps/ernie_sat/synthesize.py
+1
-2
paddlespeech/t2s/exps/ernie_sat/train.py
paddlespeech/t2s/exps/ernie_sat/train.py
+1
-2
未找到文件。
examples/aishell3/ernie_sat/local/train.sh
浏览文件 @
97965f4c
...
...
@@ -8,5 +8,5 @@ python3 ${BIN_DIR}/train.py \
--dev-metadata
=
dump/dev/norm/metadata.jsonl
\
--config
=
${
config_path
}
\
--output-dir
=
${
train_output_path
}
\
--ngpu
=
1
\
--ngpu
=
2
\
--phones-dict
=
dump/phone_id_map.txt
\ No newline at end of file
examples/aishell3_vctk/ernie_sat/local/train.sh
浏览文件 @
97965f4c
...
...
@@ -8,5 +8,5 @@ python3 ${BIN_DIR}/train.py \
--dev-metadata
=
dump/dev/norm/metadata.jsonl
\
--config
=
${
config_path
}
\
--output-dir
=
${
train_output_path
}
\
--ngpu
=
1
\
--ngpu
=
2
\
--phones-dict
=
dump/phone_id_map.txt
\ No newline at end of file
examples/vctk/ernie_sat/conf/default.yaml
浏览文件 @
97965f4c
...
...
@@ -79,7 +79,7 @@ grad_clip: 1.0
###########################################################
# TRAINING SETTING #
###########################################################
max_epoch
:
2
00
max_epoch
:
6
00
num_snapshots
:
5
###########################################################
...
...
@@ -160,4 +160,4 @@ token_list:
-
UH0
-
AW0
-
OY0
-
<sos/eos>
\ No newline at end of file
-
<sos/eos>
examples/vctk/ernie_sat/local/train.sh
浏览文件 @
97965f4c
...
...
@@ -8,5 +8,5 @@ python3 ${BIN_DIR}/train.py \
--dev-metadata
=
dump/dev/norm/metadata.jsonl
\
--config
=
${
config_path
}
\
--output-dir
=
${
train_output_path
}
\
--ngpu
=
1
\
--ngpu
=
2
\
--phones-dict
=
dump/phone_id_map.txt
\ No newline at end of file
paddlespeech/t2s/datasets/am_batch_fn.py
浏览文件 @
97965f4c
...
...
@@ -29,20 +29,13 @@ from paddlespeech.t2s.modules.nets_utils import phones_text_masking
# 因为要传参数,所以需要额外构建
def
build_erniesat_collate_fn
(
mlm_prob
:
float
=
0.8
,
mean_phn_span
:
int
=
8
,
seg_emb
:
bool
=
False
,
text_masking
:
bool
=
False
,
epoch
:
int
=-
1
,
):
if
epoch
==
-
1
:
mlm_prob_factor
=
1
else
:
mlm_prob_factor
=
0.8
def
build_erniesat_collate_fn
(
mlm_prob
:
float
=
0.8
,
mean_phn_span
:
int
=
8
,
seg_emb
:
bool
=
False
,
text_masking
:
bool
=
False
):
return
ErnieSATCollateFn
(
mlm_prob
=
mlm_prob
*
mlm_prob_factor
,
mlm_prob
=
mlm_prob
,
mean_phn_span
=
mean_phn_span
,
seg_emb
=
seg_emb
,
text_masking
=
text_masking
)
...
...
paddlespeech/t2s/exps/ernie_sat/synthesize.py
浏览文件 @
97965f4c
...
...
@@ -73,8 +73,7 @@ def evaluate(args):
mlm_prob
=
erniesat_config
.
mlm_prob
,
mean_phn_span
=
erniesat_config
.
mean_phn_span
,
seg_emb
=
erniesat_config
.
model
[
'enc_input_layer'
]
==
'sega_mlm'
,
text_masking
=
False
,
epoch
=-
1
)
text_masking
=
False
)
gen_raw
=
True
erniesat_mu
,
erniesat_std
=
np
.
load
(
args
.
erniesat_stat
)
...
...
paddlespeech/t2s/exps/ernie_sat/train.py
浏览文件 @
97965f4c
...
...
@@ -84,8 +84,7 @@ def train_sp(args, config):
mlm_prob
=
config
.
mlm_prob
,
mean_phn_span
=
config
.
mean_phn_span
,
seg_emb
=
config
.
model
[
'enc_input_layer'
]
==
'sega_mlm'
,
text_masking
=
config
[
"model"
][
"text_masking"
],
epoch
=
config
[
"max_epoch"
])
text_masking
=
config
[
"model"
][
"text_masking"
])
train_sampler
=
DistributedBatchSampler
(
train_dataset
,
...
...
编辑
预览
Markdown
is supported
0%
请重试
或
添加新附件
.
添加附件
取消
You are about to add
0
people
to the discussion. Proceed with caution.
先完成此消息的编辑!
取消
想要评论请
注册
或
登录