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
2c75c923
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
2c75c923
编写于
8月 23, 2021
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_mfa
上级
9ac6d65a
变更
7
隐藏空白更改
内联
并排
Showing
7 changed file
with
79 addition
and
48 deletion
+79
-48
deepspeech/decoders/swig/setup.py
deepspeech/decoders/swig/setup.py
+3
-2
examples/aishell/s0/conf/augmentation.json
examples/aishell/s0/conf/augmentation.json
+1
-1
examples/thchs30/a0/local/data.sh
examples/thchs30/a0/local/data.sh
+22
-16
examples/thchs30/a0/local/gen_word2phone.py
examples/thchs30/a0/local/gen_word2phone.py
+38
-18
examples/thchs30/a0/local/reorganize_thchs30.py
examples/thchs30/a0/local/reorganize_thchs30.py
+5
-4
examples/thchs30/a0/run.sh
examples/thchs30/a0/run.sh
+9
-6
tools/extras/install_mfa.sh
tools/extras/install_mfa.sh
+1
-1
未找到文件。
deepspeech/decoders/swig/setup.py
浏览文件 @
2c75c923
...
@@ -84,8 +84,9 @@ FILES = glob.glob('kenlm/util/*.cc') \
...
@@ -84,8 +84,9 @@ FILES = glob.glob('kenlm/util/*.cc') \
FILES
+=
glob
.
glob
(
'openfst-1.6.3/src/lib/*.cc'
)
FILES
+=
glob
.
glob
(
'openfst-1.6.3/src/lib/*.cc'
)
FILES
=
[
FILES
=
[
fn
for
fn
in
FILES
if
not
(
fn
.
endswith
(
'main.cc'
)
or
fn
.
endswith
(
'test.cc'
)
fn
for
fn
in
FILES
or
fn
.
endswith
(
'unittest.cc'
))
if
not
(
fn
.
endswith
(
'main.cc'
)
or
fn
.
endswith
(
'test.cc'
)
or
fn
.
endswith
(
'unittest.cc'
))
]
]
LIBS
=
[
'stdc++'
]
LIBS
=
[
'stdc++'
]
...
...
examples/aishell/s0/conf/augmentation.json
浏览文件 @
2c75c923
...
@@ -33,4 +33,4 @@
...
@@ -33,4 +33,4 @@
},
},
"prob"
:
1.0
"prob"
:
1.0
}
}
]
]
\ No newline at end of file
examples/thchs30/a0/local/data.sh
浏览文件 @
2c75c923
...
@@ -20,27 +20,33 @@ if [ ${stage} -le -1 ] && [ ${stop_stage} -ge -1 ]; then
...
@@ -20,27 +20,33 @@ if [ ${stage} -le -1 ] && [ ${stop_stage} -ge -1 ]; then
echo
"Prepare THCHS-30 failed. Terminated."
echo
"Prepare THCHS-30 failed. Terminated."
exit
1
exit
1
fi
fi
fi
fi
# dump manifest to data/
if
[
${
stage
}
-le
0
]
&&
[
${
stop_stage
}
-ge
0
]
;
then
python3
${
MAIN_ROOT
}
/utils/dump_manifest.py
--manifest-path
=
data/manifest.train
--output-dir
=
data
# dump manifest to data/
python3
${
MAIN_ROOT
}
/utils/dump_manifest.py
--manifest-path
=
data/manifest.train
--output-dir
=
data
# copy files to data/dict to gen word.lexicon
fi
cp
${
TARGET_DIR
}
/thchs30/data_thchs30/lm_word/lexicon.txt data/dict/lm_word_lexicon_1
cp
${
TARGET_DIR
}
/thchs30/resource/dict/lexicon.txt data/dict/lm_word_lexicon_2
# copy phone.lexicon to data/dict
if
[
${
stage
}
-le
1
]
&&
[
${
stop_stage
}
-ge
1
]
;
then
cp
${
TARGET_DIR
}
/thchs30/data_thchs30/lm_phone/lexicon.txt data/dict/phone.lexicon
# copy files to data/dict to gen word.lexicon
cp
${
TARGET_DIR
}
/thchs30/data_thchs30/lm_word/lexicon.txt data/dict/lm_word_lexicon_1
cp
${
TARGET_DIR
}
/thchs30/resource/dict/lexicon.txt data/dict/lm_word_lexicon_2
# copy phone.lexicon to data/dict
cp
${
TARGET_DIR
}
/thchs30/data_thchs30/lm_phone/lexicon.txt data/dict/phone.lexicon
fi
# gen word.lexicon
if
[
${
stage
}
-le
2
]
&&
[
${
stop_stage
}
-ge
2
]
;
then
python
local
/gen_word2phone.py
--root-dir
=
data/dict
--output-dir
=
data/dict
# gen word.lexicon
python
local
/gen_word2phone.py
--lexicon-files
=
"data/dict/lm_word_lexicon_1 data/dict/lm_word_lexicon_2"
--output-path
=
data/dict/word.lexicon
fi
# reorganize dataset for MFA
if
[
${
stage
}
-le
3
]
&&
[
${
stop_stage
}
-ge
3
]
;
then
if
[
!
-d
$EXP_DIR
/thchs30_corpus
]
;
then
# reorganize dataset for MFA
echo
"reorganizing thchs30 corpus..."
if
[
!
-d
$EXP_DIR
/thchs30_corpus
]
;
then
python
local
/reorganize_thchs30.py
--root-dir
=
data
--output-dir
=
data/thchs30_corpus
--script-type
=
$LEXICON_NAME
echo
"reorganizing thchs30 corpus..."
echo
"reorganization done."
python
local
/reorganize_thchs30.py
--root-dir
=
data
--output-dir
=
data/thchs30_corpus
--script-type
=
$LEXICON_NAME
echo
"reorganization done."
fi
fi
fi
echo
"THCHS-30 data preparation done."
echo
"THCHS-30 data preparation done."
...
...
examples/thchs30/a0/local/gen_word2phone.py
浏览文件 @
2c75c923
...
@@ -18,6 +18,7 @@ file2: THCHS-30/resource/dict/lexicon.txt
...
@@ -18,6 +18,7 @@ file2: THCHS-30/resource/dict/lexicon.txt
import
argparse
import
argparse
from
collections
import
defaultdict
from
collections
import
defaultdict
from
pathlib
import
Path
from
pathlib
import
Path
from
typing
import
List
from
typing
import
Union
from
typing
import
Union
# key: (cn, ('ee', 'er4')),value: count
# key: (cn, ('ee', 'er4')),value: count
...
@@ -34,7 +35,7 @@ def is_Chinese(ch):
...
@@ -34,7 +35,7 @@ def is_Chinese(ch):
return
False
return
False
def
proc_line
(
line
):
def
proc_line
(
line
:
str
):
line
=
line
.
strip
()
line
=
line
.
strip
()
if
is_Chinese
(
line
[
0
]):
if
is_Chinese
(
line
[
0
]):
line_list
=
line
.
split
()
line_list
=
line
.
split
()
...
@@ -49,20 +50,25 @@ def proc_line(line):
...
@@ -49,20 +50,25 @@ def proc_line(line):
cn_phones_counter
[(
cn
,
phones
)]
+=
1
cn_phones_counter
[(
cn
,
phones
)]
+=
1
def
gen_lexicon
(
root_dir
:
Union
[
str
,
Path
],
output_dir
:
Union
[
str
,
Path
]):
"""
root_dir
=
Path
(
root_dir
).
expanduser
()
example lines of output
output_dir
=
Path
(
output_dir
).
expanduser
()
the first column is a Chinese character
output_dir
.
mkdir
(
parents
=
True
,
exist_ok
=
True
)
the second is the probability of this pronunciation
file1
=
root_dir
/
"lm_word_lexicon_1"
and the rest are the phones of this pronunciation
file2
=
root_dir
/
"lm_word_lexicon_2"
一 0.22 ii i1↩
write_file
=
output_dir
/
"word.lexicon"
一 0.45 ii i4↩
一 0.32 ii i2↩
一 0.01 ii i5
"""
def
gen_lexicon
(
lexicon_files
:
List
[
Union
[
str
,
Path
]],
output_path
:
Union
[
str
,
Path
]):
for
file_path
in
lexicon_files
:
with
open
(
file_path
,
"r"
)
as
f1
:
for
line
in
f1
:
proc_line
(
line
)
with
open
(
file1
,
"r"
)
as
f1
:
for
line
in
f1
:
proc_line
(
line
)
with
open
(
file2
,
"r"
)
as
f2
:
for
line
in
f2
:
proc_line
(
line
)
for
key
in
cn_phones_counter
:
for
key
in
cn_phones_counter
:
cn
=
key
[
0
]
cn
=
key
[
0
]
cn_counter
[
cn
].
append
((
key
[
1
],
cn_phones_counter
[
key
]))
cn_counter
[
cn
].
append
((
key
[
1
],
cn_phones_counter
[
key
]))
...
@@ -75,7 +81,8 @@ def gen_lexicon(root_dir: Union[str, Path], output_dir: Union[str, Path]):
...
@@ -75,7 +81,8 @@ def gen_lexicon(root_dir: Union[str, Path], output_dir: Union[str, Path]):
p
=
round
(
p
,
2
)
p
=
round
(
p
,
2
)
if
p
>
0
:
if
p
>
0
:
cn_counter_p
[
key
].
append
((
item
[
0
],
p
))
cn_counter_p
[
key
].
append
((
item
[
0
],
p
))
with
open
(
write_file
,
"w"
)
as
wf
:
with
open
(
output_path
,
"w"
)
as
wf
:
for
key
in
cn_counter_p
:
for
key
in
cn_counter_p
:
phone_p_list
=
cn_counter_p
[
key
]
phone_p_list
=
cn_counter_p
[
key
]
for
item
in
phone_p_list
:
for
item
in
phone_p_list
:
...
@@ -87,8 +94,21 @@ if __name__ == "__main__":
...
@@ -87,8 +94,21 @@ if __name__ == "__main__":
parser
=
argparse
.
ArgumentParser
(
parser
=
argparse
.
ArgumentParser
(
description
=
"Gen Chinese characters to phone lexicon for THCHS-30 dataset"
description
=
"Gen Chinese characters to phone lexicon for THCHS-30 dataset"
)
)
# A line of word_lexicon:
# 一丁点 ii i4 d ing1 d ian3
# the first is word, and the rest are the phones of the word, and the len of phones is twice of the word's len
parser
.
add_argument
(
"--lexicon-files"
,
type
=
str
,
default
=
"data/dict/lm_word_lexicon_1 data/dict/lm_word_lexicon_2"
,
help
=
"lm_word_lexicon files"
)
parser
.
add_argument
(
parser
.
add_argument
(
"--root-dir"
,
type
=
str
,
help
=
"dir to thchs30 lm_word_lexicons"
)
"--output-path"
,
parser
.
add_argument
(
"--output-dir"
,
type
=
str
,
help
=
"path to save outputs"
)
type
=
str
,
default
=
"data/dict/word.lexicon"
,
help
=
"path to save output word2phone lexicon"
)
args
=
parser
.
parse_args
()
args
=
parser
.
parse_args
()
gen_lexicon
(
args
.
root_dir
,
args
.
output_dir
)
lexicon_files
=
args
.
lexicon_files
.
split
(
" "
)
output_path
=
Path
(
args
.
output_path
).
expanduser
()
gen_lexicon
(
lexicon_files
,
output_path
)
examples/thchs30/a0/local/reorganize_thchs30.py
浏览文件 @
2c75c923
...
@@ -58,8 +58,6 @@ def write_lab(root_dir: Union[str, Path],
...
@@ -58,8 +58,6 @@ def write_lab(root_dir: Union[str, Path],
def
reorganize_thchs30
(
root_dir
:
Union
[
str
,
Path
],
def
reorganize_thchs30
(
root_dir
:
Union
[
str
,
Path
],
output_dir
:
Union
[
str
,
Path
]
=
None
,
output_dir
:
Union
[
str
,
Path
]
=
None
,
script_type
=
'phone'
):
script_type
=
'phone'
):
root_dir
=
Path
(
root_dir
).
expanduser
()
output_dir
=
Path
(
output_dir
).
expanduser
()
output_dir
.
mkdir
(
parents
=
True
,
exist_ok
=
True
)
output_dir
.
mkdir
(
parents
=
True
,
exist_ok
=
True
)
link_wav
(
root_dir
,
output_dir
)
link_wav
(
root_dir
,
output_dir
)
write_lab
(
root_dir
,
output_dir
,
script_type
)
write_lab
(
root_dir
,
output_dir
,
script_type
)
...
@@ -72,12 +70,15 @@ if __name__ == "__main__":
...
@@ -72,12 +70,15 @@ if __name__ == "__main__":
parser
.
add_argument
(
parser
.
add_argument
(
"--output-dir"
,
"--output-dir"
,
type
=
str
,
type
=
str
,
help
=
"path to save outputs(audio and transcriptions)"
)
help
=
"path to save outputs
(audio and transcriptions)"
)
parser
.
add_argument
(
parser
.
add_argument
(
"--script-type"
,
"--script-type"
,
type
=
str
,
type
=
str
,
default
=
"phone"
,
default
=
"phone"
,
help
=
"type of lab ('word'/'syllable'/'phone')"
)
help
=
"type of lab ('word'/'syllable'/'phone')"
)
args
=
parser
.
parse_args
()
args
=
parser
.
parse_args
()
reorganize_thchs30
(
args
.
root_dir
,
args
.
output_dir
,
args
.
script_type
)
root_dir
=
Path
(
args
.
root_dir
).
expanduser
()
output_dir
=
Path
(
args
.
output_dir
).
expanduser
()
reorganize_thchs30
(
root_dir
,
output_dir
,
args
.
script_type
)
examples/thchs30/a0/run.sh
浏览文件 @
2c75c923
...
@@ -14,14 +14,17 @@ source ${MAIN_ROOT}/utils/parse_options.sh || exit 1;
...
@@ -14,14 +14,17 @@ source ${MAIN_ROOT}/utils/parse_options.sh || exit 1;
# gen lexicon relink gen dump
# gen lexicon relink gen dump
if
[
${
stage
}
-le
0
]
&&
[
${
stop_stage
}
-ge
0
]
;
then
if
[
${
stage
}
-le
0
]
&&
[
${
stop_stage
}
-ge
0
]
;
then
# prepare data
# prepare data
bash ./local/data.sh
$LEXICON_NAME
||
exit
-1
echo
"Start prepare thchs30 data for MFA ..."
bash ./local/data.sh
$LEXICON_NAME
||
exit
-1
fi
fi
# run MFA
if
[
${
stage
}
-le
1
]
&&
[
${
stop_stage
}
-ge
1
]
;
then
if
[
!
-d
"
$EXP_DIR
/thchs30_alignment"
]
;
then
# run MFA
echo
"Start MFA training..."
if
[
!
-d
"
$EXP_DIR
/thchs30_alignment"
]
;
then
mfa_train_and_align data/thchs30_corpus data/dict/
$LEXICON_NAME
.lexicon
$EXP_DIR
/thchs30_alignment
-o
$EXP_DIR
/thchs30_model
--clean
--verbose
--temp_directory
exp/.mfa_train_and_align
--num_jobs
$NUM_JOBS
echo
"Start MFA training ..."
echo
"training done!
\n
results:
$EXP_DIR
/thchs30_alignment
\n
model:
$EXP_DIR
/thchs30_model
\n
"
mfa_train_and_align data/thchs30_corpus data/dict/
$LEXICON_NAME
.lexicon
$EXP_DIR
/thchs30_alignment
-o
$EXP_DIR
/thchs30_model
--clean
--verbose
--temp_directory
exp/.mfa_train_and_align
--num_jobs
$NUM_JOBS
echo
"MFA training done!
\n
results:
$EXP_DIR
/thchs30_alignment
\n
model:
$EXP_DIR
/thchs30_model
\n
"
fi
fi
fi
...
...
tools/extras/install_mfa.sh
浏览文件 @
2c75c923
...
@@ -4,7 +4,7 @@
...
@@ -4,7 +4,7 @@
test
-d
Montreal-Forced-Aligner
||
git clone https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner.git
test
-d
Montreal-Forced-Aligner
||
git clone https://github.com/MontrealCorpusTools/Montreal-Forced-Aligner.git
pushd
Montreal-Forced-Aligner
&&
git checkout v2.0.0a7
&&
python setup.py
install
pushd
Montreal-Forced-Aligner
&&
python setup.py
install
&&
popd
test
-d
kaldi
||
{
echo
"need install kaldi first"
;
exit
1
;
}
test
-d
kaldi
||
{
echo
"need install kaldi first"
;
exit
1
;
}
...
...
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