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
Parakeet
提交
b604d1c7
P
Parakeet
项目概览
PaddlePaddle
/
Parakeet
通知
8
Star
3
Fork
0
代码
文件
提交
分支
Tags
贡献者
分支图
Diff
Issue
19
列表
看板
标记
里程碑
合并请求
0
Wiki
0
Wiki
分析
仓库
DevOps
项目成员
Pages
P
Parakeet
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
19
Issue
19
列表
看板
标记
里程碑
合并请求
0
合并请求
0
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
b604d1c7
编写于
8月 12, 2020
作者:
L
liuyibing01
浏览文件
操作
浏览文件
下载
差异文件
Merge branch 'develop' into 'master'
examples/deepvoice3 minor fixes See merge request !73
上级
610181d4
3717ac13
变更
4
隐藏空白更改
内联
并排
Showing
4 changed file
with
5 addition
and
6 deletion
+5
-6
examples/deepvoice3/README.md
examples/deepvoice3/README.md
+1
-1
examples/deepvoice3/configs/ljspeech.yaml
examples/deepvoice3/configs/ljspeech.yaml
+1
-0
examples/deepvoice3/data.py
examples/deepvoice3/data.py
+2
-4
examples/deepvoice3/train.py
examples/deepvoice3/train.py
+1
-1
未找到文件。
examples/deepvoice3/README.md
浏览文件 @
b604d1c7
...
@@ -87,7 +87,7 @@ runs/Jul07_09-39-34_instance-mqcyj27y-4/
...
@@ -87,7 +87,7 @@ runs/Jul07_09-39-34_instance-mqcyj27y-4/
...
...
```
```
Since e use waveflow to synthesize audio while training, so download the trained waveflow model and extract it in current directory before training.
Since
w
e use waveflow to synthesize audio while training, so download the trained waveflow model and extract it in current directory before training.
```
bash
```
bash
wget https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_res128_ljspeech_ckpt_1.0.zip
wget https://paddlespeech.bj.bcebos.com/Parakeet/waveflow_res128_ljspeech_ckpt_1.0.zip
...
...
examples/deepvoice3/configs/ljspeech.yaml
浏览文件 @
b604d1c7
...
@@ -39,6 +39,7 @@ clip_value: 5.0
...
@@ -39,6 +39,7 @@ clip_value: 5.0
clip_norm
:
100.0
clip_norm
:
100.0
# training:
# training:
max_iteration
:
1000000
batch_size
:
16
batch_size
:
16
report_interval
:
10000
report_interval
:
10000
save_interval
:
10000
save_interval
:
10000
...
...
examples/deepvoice3/data.py
浏览文件 @
b604d1c7
...
@@ -62,10 +62,8 @@ class DataCollector(object):
...
@@ -62,10 +62,8 @@ class DataCollector(object):
for
example
in
examples
:
for
example
in
examples
:
text
,
spec
,
mel
,
_
=
example
text
,
spec
,
mel
,
_
=
example
text_seqs
.
append
(
en
.
text_to_sequence
(
text
,
self
.
p_pronunciation
))
text_seqs
.
append
(
en
.
text_to_sequence
(
text
,
self
.
p_pronunciation
))
# if max_frames - mel.shape[0] < 0:
specs
.
append
(
np
.
pad
(
spec
,
[(
0
,
max_frames
-
spec
.
shape
[
0
]),
(
0
,
0
)],
mode
=
"constant"
))
# import pdb; pdb.set_trace()
mels
.
append
(
np
.
pad
(
mel
,
[(
0
,
max_frames
-
mel
.
shape
[
0
]),
(
0
,
0
)],
mode
=
"constant"
))
specs
.
append
(
np
.
pad
(
spec
,
[(
0
,
max_frames
-
spec
.
shape
[
0
]),
(
0
,
0
)]))
mels
.
append
(
np
.
pad
(
mel
,
[(
0
,
max_frames
-
mel
.
shape
[
0
]),
(
0
,
0
)]))
specs
=
np
.
stack
(
specs
)
specs
=
np
.
stack
(
specs
)
mels
=
np
.
stack
(
mels
)
mels
=
np
.
stack
(
mels
)
...
...
examples/deepvoice3/train.py
浏览文件 @
b604d1c7
...
@@ -81,7 +81,7 @@ def train(args, config):
...
@@ -81,7 +81,7 @@ def train(args, config):
optim
=
create_optimizer
(
model
,
config
)
optim
=
create_optimizer
(
model
,
config
)
global
global_step
global
global_step
max_iteration
=
1000000
max_iteration
=
config
[
"max_iteration"
]
iterator
=
iter
(
tqdm
.
tqdm
(
train_loader
))
iterator
=
iter
(
tqdm
.
tqdm
(
train_loader
))
while
global_step
<=
max_iteration
:
while
global_step
<=
max_iteration
:
...
...
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