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
ac370eca
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
ac370eca
编写于
6月 07, 2017
作者:
Y
Yibing Liu
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
add annotations
上级
51f35a53
变更
2
隐藏空白更改
内联
并排
Showing
2 changed file
with
8 addition
and
8 deletion
+8
-8
decoder.py
decoder.py
+5
-6
infer.py
infer.py
+3
-2
未找到文件。
decoder.py
浏览文件 @
ac370eca
...
@@ -68,9 +68,9 @@ class Scorer(object):
...
@@ -68,9 +68,9 @@ class Scorer(object):
# execute evaluation
# execute evaluation
def
evaluate
(
self
,
sentence
,
bos
=
True
,
eos
=
False
):
def
evaluate
(
self
,
sentence
,
bos
=
True
,
eos
=
False
):
lm
=
self
.
language_model_score
(
sentence
,
bos
,
eos
)
lm
=
self
.
language_model_score
(
sentence
,
bos
,
eos
)
word_c
ou
nt
=
self
.
word_count
(
sentence
)
word_cnt
=
self
.
word_count
(
sentence
)
score
=
np
.
power
(
lm
,
self
.
_alpha
)
\
score
=
np
.
power
(
lm
,
self
.
_alpha
)
\
*
np
.
power
(
word_c
ou
nt
,
self
.
_beta
)
*
np
.
power
(
word_cnt
,
self
.
_beta
)
return
score
return
score
...
@@ -104,19 +104,18 @@ def ctc_beam_search_decoder(probs_seq,
...
@@ -104,19 +104,18 @@ def ctc_beam_search_decoder(probs_seq,
:rtype: list
:rtype: list
'''
'''
# dimension check
for
prob_list
in
probs_seq
:
for
prob_list
in
probs_seq
:
if
not
len
(
prob_list
)
==
len
(
vocabulary
)
+
1
:
if
not
len
(
prob_list
)
==
len
(
vocabulary
)
+
1
:
raise
ValueError
(
"probs dimension mismatchedd with vocabulary"
)
raise
ValueError
(
"probs dimension mismatchedd with vocabulary"
)
max_time_steps
=
len
(
probs_seq
)
max_time_steps
=
len
(
probs_seq
)
if
not
max_time_steps
>
0
:
raise
ValueError
(
"probs_seq shouldn't be empty"
)
# blank_id check
probs_dim
=
len
(
probs_seq
[
0
])
probs_dim
=
len
(
probs_seq
[
0
])
if
not
blank_id
<
probs_dim
:
if
not
blank_id
<
probs_dim
:
raise
ValueError
(
"blank_id shouldn't be greater than probs dimension"
)
raise
ValueError
(
"blank_id shouldn't be greater than probs dimension"
)
# assign space_id
if
' '
not
in
vocabulary
:
if
' '
not
in
vocabulary
:
raise
ValueError
(
"space doesn't exist in vocabulary"
)
raise
ValueError
(
"space doesn't exist in vocabulary"
)
space_id
=
vocabulary
.
index
(
' '
)
space_id
=
vocabulary
.
index
(
' '
)
...
...
infer.py
浏览文件 @
ac370eca
...
@@ -77,7 +77,7 @@ parser.add_argument(
...
@@ -77,7 +77,7 @@ parser.add_argument(
"--language_model_path"
,
"--language_model_path"
,
default
=
"./data/1Billion.klm"
,
default
=
"./data/1Billion.klm"
,
type
=
str
,
type
=
str
,
help
=
"Path for language model. (default: %(default)
d
)"
)
help
=
"Path for language model. (default: %(default)
s
)"
)
parser
.
add_argument
(
parser
.
add_argument
(
"--alpha"
,
"--alpha"
,
default
=
0.0
,
default
=
0.0
,
...
@@ -93,7 +93,7 @@ args = parser.parse_args()
...
@@ -93,7 +93,7 @@ args = parser.parse_args()
def
infer
():
def
infer
():
"""
"""
Max-ctc-decoding
for DeepSpeech2.
Inference
for DeepSpeech2.
"""
"""
# initialize data generator
# initialize data generator
data_generator
=
DataGenerator
(
data_generator
=
DataGenerator
(
...
@@ -174,6 +174,7 @@ def infer():
...
@@ -174,6 +174,7 @@ def infer():
print
(
"
\n
Target Transcription:
\t
%s"
%
target_transcription
)
print
(
"
\n
Target Transcription:
\t
%s"
%
target_transcription
)
for
index
in
range
(
args
.
num_results_per_sample
):
for
index
in
range
(
args
.
num_results_per_sample
):
result
=
beam_search_result
[
index
]
result
=
beam_search_result
[
index
]
#output: index, log prob, beam result
print
(
"Beam %d: %f
\t
%s"
%
(
index
,
result
[
0
],
result
[
1
]))
print
(
"Beam %d: %f
\t
%s"
%
(
index
,
result
[
0
],
result
[
1
]))
else
:
else
:
raise
ValueError
(
"Decoding method [%s] is not supported."
%
method
)
raise
ValueError
(
"Decoding method [%s] is not supported."
%
method
)
...
...
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