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
3222b2c5
D
DeepSpeech
项目概览
PaddlePaddle
/
DeepSpeech
大约 1 年 前同步成功
通知
206
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
体验新版 GitCode,发现更多精彩内容 >>
提交
3222b2c5
编写于
1月 28, 2022
作者:
W
WilliamZhang06
浏览文件
操作
浏览文件
下载
电子邮件补丁
差异文件
added asr engine, test=doc
上级
6d810d2a
变更
4
隐藏空白更改
内联
并排
Showing
4 changed file
with
164 addition
and
19 deletion
+164
-19
speechserving/speechserving/conf/asr/asr.yaml
speechserving/speechserving/conf/asr/asr.yaml
+3
-0
speechserving/speechserving/engine/asr/python/asr_engine.py
speechserving/speechserving/engine/asr/python/asr_engine.py
+153
-10
speechserving/speechserving/restful/asr_api.py
speechserving/speechserving/restful/asr_api.py
+3
-4
speechserving/tests/http_client.py
speechserving/tests/http_client.py
+5
-5
未找到文件。
speechserving/speechserving/conf/asr/asr.yaml
浏览文件 @
3222b2c5
model
:
'
conformer_wenetspeech'
lang
:
'
zh'
sample_rate
:
16000
cfg_path
:
"
/home/users/zhangyinhui/.paddlespeech/models/conformer_wenetspeech-zh-16k/asr1_conformer_wenetspeech_ckpt_0.1.1.model.tar/model.yaml"
ckpt_path
:
"
/home/users/zhangyinhui/.paddlespeech/models/conformer_wenetspeech-zh-16k/asr1_conformer_wenetspeech_ckpt_0.1.1.model.tar/exp/conformer/checkpoints/wenetspeech"
decode_method
:
'
attention_rescoring'
force_yes
:
False
speechserving/speechserving/engine/asr/python/asr_engine.py
浏览文件 @
3222b2c5
...
...
@@ -11,29 +11,172 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from
engine.base_engine
import
BaseEngine
import
paddle
import
io
import
soundfile
import
os
import
librosa
from
typing
import
List
from
typing
import
Optional
from
typing
import
Union
from
paddlespeech.cli.log
import
logger
from
paddlespeech.cli.asr.infer
import
ASRExecutor
from
paddlespeech.s2t.frontend.featurizer.text_featurizer
import
TextFeaturizer
from
paddlespeech.s2t.transform.transformation
import
Transformation
from
paddlespeech.s2t.utils.dynamic_import
import
dynamic_import
from
paddlespeech.s2t.utils.utility
import
UpdateConfig
from
utils.log
import
logger
from
engine.base_engine
import
BaseEngine
from
utils.config
import
get_config
__all__
=
[
'ASREngine'
]
class
ASRServerExecutor
(
ASRExecutor
):
def
__init__
(
self
):
super
().
__init__
()
pass
def
_check
(
self
,
audio_file
:
str
,
sample_rate
:
int
,
force_yes
:
bool
):
self
.
sample_rate
=
sample_rate
if
self
.
sample_rate
!=
16000
and
self
.
sample_rate
!=
8000
:
logger
.
error
(
"please input --sr 8000 or --sr 16000"
)
return
False
logger
.
info
(
"checking the audio file format......"
)
try
:
audio
,
audio_sample_rate
=
soundfile
.
read
(
audio_file
,
dtype
=
"int16"
,
always_2d
=
True
)
except
Exception
as
e
:
logger
.
exception
(
e
)
logger
.
error
(
"can not open the audio file, please check the audio file format is 'wav'.
\n
\
you can try to use sox to change the file format.
\n
\
For example:
\n
\
sample rate: 16k
\n
\
sox input_audio.xx --rate 16k --bits 16 --channels 1 output_audio.wav
\n
\
sample rate: 8k
\n
\
sox input_audio.xx --rate 8k --bits 16 --channels 1 output_audio.wav
\n
\
"
)
logger
.
info
(
"The sample rate is %d"
%
audio_sample_rate
)
if
audio_sample_rate
!=
self
.
sample_rate
:
logger
.
warning
(
"The sample rate of the input file is not {}.
\n
\
The program will resample the wav file to {}.
\n
\
If the result does not meet your expectations,
\n
\
Please input the 16k 16 bit 1 channel wav file.
\
"
.
format
(
self
.
sample_rate
,
self
.
sample_rate
))
self
.
change_format
=
True
else
:
logger
.
info
(
"The audio file format is right"
)
self
.
change_format
=
False
return
True
def
preprocess
(
self
,
model_type
:
str
,
input
:
Union
[
str
,
os
.
PathLike
]):
"""
Input preprocess and return paddle.Tensor stored in self.input.
Input content can be a text(tts), a file(asr, cls) or a streaming(not supported yet).
"""
audio_file
=
input
# logger.info("Preprocess audio_file:" + audio_file)
# Get the object for feature extraction
if
"deepspeech2online"
in
model_type
or
"deepspeech2offline"
in
model_type
:
audio
,
_
=
self
.
collate_fn_test
.
process_utterance
(
audio_file
=
audio_file
,
transcript
=
" "
)
audio_len
=
audio
.
shape
[
0
]
audio
=
paddle
.
to_tensor
(
audio
,
dtype
=
'float32'
)
audio_len
=
paddle
.
to_tensor
(
audio_len
)
audio
=
paddle
.
unsqueeze
(
audio
,
axis
=
0
)
# vocab_list = collate_fn_test.vocab_list
self
.
_inputs
[
"audio"
]
=
audio
self
.
_inputs
[
"audio_len"
]
=
audio_len
logger
.
info
(
f
"audio feat shape:
{
audio
.
shape
}
"
)
elif
"conformer"
in
model_type
or
"transformer"
in
model_type
or
"wenetspeech"
in
model_type
:
logger
.
info
(
"get the preprocess conf"
)
preprocess_conf
=
self
.
config
.
preprocess_config
preprocess_args
=
{
"train"
:
False
}
preprocessing
=
Transformation
(
preprocess_conf
)
logger
.
info
(
"read the audio file"
)
audio
,
audio_sample_rate
=
soundfile
.
read
(
audio_file
,
dtype
=
"int16"
,
always_2d
=
True
)
if
self
.
change_format
:
if
audio
.
shape
[
1
]
>=
2
:
audio
=
audio
.
mean
(
axis
=
1
,
dtype
=
np
.
int16
)
else
:
audio
=
audio
[:,
0
]
# pcm16 -> pcm 32
audio
=
self
.
_pcm16to32
(
audio
)
audio
=
librosa
.
resample
(
audio
,
audio_sample_rate
,
self
.
sample_rate
)
audio_sample_rate
=
self
.
sample_rate
# pcm32 -> pcm 16
audio
=
self
.
_pcm32to16
(
audio
)
else
:
audio
=
audio
[:,
0
]
logger
.
info
(
f
"audio shape:
{
audio
.
shape
}
"
)
# fbank
audio
=
preprocessing
(
audio
,
**
preprocess_args
)
audio_len
=
paddle
.
to_tensor
(
audio
.
shape
[
0
])
audio
=
paddle
.
to_tensor
(
audio
,
dtype
=
'float32'
).
unsqueeze
(
axis
=
0
)
self
.
_inputs
[
"audio"
]
=
audio
self
.
_inputs
[
"audio_len"
]
=
audio_len
logger
.
info
(
f
"audio feat shape:
{
audio
.
shape
}
"
)
else
:
raise
Exception
(
"wrong type"
)
class
ASREngine
(
BaseEngine
):
"""ASR server engine
Args:
metaclass: Defaults to Singleton.
"""
def
__init__
(
self
):
super
(
ASREngine
,
self
).
__init__
()
def
init
(
self
,
config_file
:
str
):
self
.
config_file
=
config_file
self
.
executor
=
None
self
.
executor
=
ASRServerExecutor
()
self
.
config
=
get_config
(
config_file
)
paddle
.
set_device
(
paddle
.
get_device
())
self
.
executor
.
_init_from_path
(
self
.
config
.
model
,
self
.
config
.
lang
,
self
.
config
.
sample_rate
,
self
.
config
.
cfg_path
,
self
.
config
.
decode_method
,
self
.
config
.
ckpt_path
)
logger
.
info
(
"Initialize ASR server engine successfully."
)
self
.
input
=
None
self
.
output
=
None
config
=
get_config
(
self
.
config_file
)
pass
def
postprocess
(
self
):
pass
def
run
(
self
,
audio_data
):
if
self
.
executor
.
_check
(
io
.
BytesIO
(
audio_data
),
self
.
config
.
sample_rate
,
self
.
config
.
force_yes
):
self
.
executor
.
preprocess
(
self
.
config
.
model
,
io
.
BytesIO
(
audio_data
))
self
.
executor
.
infer
(
self
.
config
.
model
)
self
.
output
=
self
.
executor
.
postprocess
()
# Retrieve result of asr.
else
:
logger
.
info
(
"file check failed!"
)
def
run
(
self
):
logger
.
info
(
"start run asr engine"
)
return
"hello world"
def
postprocess
(
self
):
return
self
.
output
speechserving/speechserving/restful/asr_api.py
浏览文件 @
3222b2c5
...
...
@@ -41,12 +41,11 @@ def asr(request_body: ASRRequest):
Returns:
json: [description]
"""
audio_data
=
base64
.
b64decode
(
request_body
.
audio
)
# single
asr_engine
=
ASREngine
()
print
(
"asr_engine id :"
,
id
(
asr_engine
))
asr_results
=
asr_engine
.
run
()
asr_engine
.
postprocess
()
asr_engine
.
run
(
audio_data
)
asr_results
=
asr_engine
.
postprocess
()
json_body
=
{
"success"
:
True
,
...
...
speechserving/tests/http_client.py
浏览文件 @
3222b2c5
...
...
@@ -14,6 +14,8 @@ import requests
import
json
import
time
import
base64
import
soundfile
import
io
import
argparse
...
...
@@ -36,11 +38,11 @@ def main(args):
# start Timestamp
time_start
=
time
.
time
()
# test_audio_dir = "test_data
/16_audio.wav"
#
audio = readwav2base64(test_audio_dir)
test_audio_dir
=
".
/16_audio.wav"
audio
=
readwav2base64
(
test_audio_dir
)
data
=
{
"audio"
:
"exSI6ICJlbiIsCgkgICAgInBvc2l0aW9uIjogImZhbHNlIgoJf"
,
"audio"
:
audio
,
"audio_format"
:
"wav"
,
"sample_rate"
:
16000
,
"lang"
:
"zh_cn"
,
...
...
@@ -55,8 +57,6 @@ def main(args):
print
(
r
.
json
())
if
__name__
==
"__main__"
:
parser
=
argparse
.
ArgumentParser
()
parser
.
add_argument
(
"--model_type"
,
action
=
"store"
,
...
...
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