Toggle navigation
Toggle navigation
此项目
正在载入...
Sign in
xuning
/
sherpaonnx
转到一个项目
Toggle navigation
项目
群组
代码片段
帮助
Toggle navigation pinning
Project
Activity
Repository
Pipelines
Graphs
Issues
0
Merge Requests
0
Wiki
Network
Create a new issue
Builds
Commits
Authored by
Fangjun Kuang
2025-07-12 19:55:37 +0800
Browse Files
Options
Browse Files
Download
Email Patches
Plain Diff
Committed by
GitHub
2025-07-12 19:55:37 +0800
Commit
b376ad3a7cf86d23e6af1a6a53db60aaa5742e51
b376ad3a
1 parent
fc2fc3d9
Add Java/Kotlin API and Android support for ten-vad (#2389)
隐藏空白字符变更
内嵌
并排对比
正在显示
36 个修改的文件
包含
396 行增加
和
47 行删除
.github/workflows/run-java-test.yaml
go-api-examples/vad-asr-paraformer/main.go
go-api-examples/vad-asr-paraformer/run.sh
go-api-examples/vad-asr-whisper/main.go
go-api-examples/vad-asr-whisper/run.sh
go-api-examples/vad-speaker-identification/main.go
go-api-examples/vad-speaker-identification/run.sh
go-api-examples/vad-spoken-language-identification/main.go
go-api-examples/vad-spoken-language-identification/run.sh
java-api-examples/README.md
java-api-examples/TenVadRemoveSilence.java
java-api-examples/run-ten-vad-remove-silence.sh
java-api-examples/run-vad-remove-slience.sh → java-api-examples/run-vad-remove-silence.sh
python-api-examples/simulate-streaming-sense-voice-microphone.py
python-api-examples/speaker-identification-with-vad-dynamic.py
python-api-examples/speaker-identification-with-vad-non-streaming-asr-alsa.py
python-api-examples/speaker-identification-with-vad-non-streaming-asr.py
python-api-examples/speaker-identification-with-vad.py
python-api-examples/vad-alsa.py
python-api-examples/vad-microphone.py
python-api-examples/vad-remove-non-speech-segments-alsa.py
python-api-examples/vad-remove-non-speech-segments-from-file.py
python-api-examples/vad-remove-non-speech-segments.py
python-api-examples/vad-with-non-streaming-asr.py
scripts/apk/build-apk-vad.sh
sherpa-onnx/csrc/keyword-spotter-transducer-impl.h
sherpa-onnx/csrc/sherpa-onnx-vad-alsa-offline-asr.cc
sherpa-onnx/csrc/sherpa-onnx-vad-alsa.cc
sherpa-onnx/csrc/sherpa-onnx-vad-microphone-offline-asr.cc
sherpa-onnx/csrc/sherpa-onnx-vad-microphone.cc
sherpa-onnx/csrc/sherpa-onnx-vad.cc
sherpa-onnx/java-api/Makefile
sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/TenVadModelConfig.java
sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/VadModelConfig.java
sherpa-onnx/jni/voice-activity-detector.cc
sherpa-onnx/kotlin-api/Vad.kt
.github/workflows/run-java-test.yaml
查看文件 @
b376ad3
...
...
@@ -237,11 +237,20 @@ jobs:
rm *.wav
rm -rf sherpa-onnx-*
-
name
:
Run java test (
VAD
remove silence)
-
name
:
Run java test (
ten-vad
remove silence)
shell
:
bash
run
:
|
cd ./java-api-examples
./run-vad-remove-slience.sh
./run-ten-vad-remove-silence.sh
rm *.onnx
ls -lh *.wav
rm *.wav
-
name
:
Run java test (silero-vad remove silence)
shell
:
bash
run
:
|
cd ./java-api-examples
./run-vad-remove-silence.sh
rm *.onnx
ls -lh *.wav
rm *.wav
...
...
go-api-examples/vad-asr-paraformer/main.go
查看文件 @
b376ad3
...
...
@@ -15,7 +15,7 @@ func main() {
config
:=
sherpa
.
VadModelConfig
{}
// Please download silero_vad.onnx from
// https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
// https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
config
.
SileroVad
.
Model
=
"./silero_vad.onnx"
config
.
SileroVad
.
Threshold
=
0.5
...
...
go-api-examples/vad-asr-paraformer/run.sh
查看文件 @
b376ad3
...
...
@@ -3,7 +3,7 @@
set
-ex
if
[
! -f ./silero_vad.onnx
]
;
then
curl -SL -O https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
curl -SL -O https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
fi
if
[
! -f ./sherpa-onnx-paraformer-trilingual-zh-cantonese-en/model.int8.onnx
]
;
then
...
...
go-api-examples/vad-asr-whisper/main.go
查看文件 @
b376ad3
...
...
@@ -15,7 +15,7 @@ func main() {
config
:=
sherpa
.
VadModelConfig
{}
// Please download silero_vad.onnx from
// https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
// https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
config
.
SileroVad
.
Model
=
"./silero_vad.onnx"
config
.
SileroVad
.
Threshold
=
0.5
...
...
go-api-examples/vad-asr-whisper/run.sh
查看文件 @
b376ad3
...
...
@@ -3,7 +3,7 @@
set
-ex
if
[
! -f ./silero_vad.onnx
]
;
then
curl -SL -O https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
curl -SL -O https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
fi
if
[
! -f ./sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx
]
;
then
...
...
go-api-examples/vad-speaker-identification/main.go
查看文件 @
b376ad3
...
...
@@ -89,7 +89,7 @@ func createVad() *sherpa.VoiceActivityDetector {
config
:=
sherpa
.
VadModelConfig
{}
// Please download silero_vad.onnx from
// https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
// https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
config
.
SileroVad
.
Model
=
"./silero_vad.onnx"
config
.
SileroVad
.
Threshold
=
0.5
...
...
go-api-examples/vad-speaker-identification/run.sh
查看文件 @
b376ad3
...
...
@@ -11,7 +11,7 @@ if [ ! -f ./sr-data/enroll/fangjun-sr-1.wav ]; then
fi
if
[
! -f ./silero_vad.onnx
]
;
then
curl -SL -O https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
curl -SL -O https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
fi
go mod tidy
...
...
go-api-examples/vad-spoken-language-identification/main.go
查看文件 @
b376ad3
...
...
@@ -15,7 +15,7 @@ func main() {
config
:=
sherpa
.
VadModelConfig
{}
// Please download silero_vad.onnx from
// https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
// https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
config
.
SileroVad
.
Model
=
"./silero_vad.onnx"
config
.
SileroVad
.
Threshold
=
0.5
...
...
go-api-examples/vad-spoken-language-identification/run.sh
查看文件 @
b376ad3
...
...
@@ -3,7 +3,7 @@
set
-ex
if
[
! -f ./silero_vad.onnx
]
;
then
curl -SL -O https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
curl -SL -O https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
fi
if
[
! -f ./sherpa-onnx-whisper-tiny/tiny-encoder.int8.onnx
]
;
then
...
...
java-api-examples/README.md
查看文件 @
b376ad3
...
...
@@ -113,6 +113,7 @@ The punctuation model supports both English and Chinese.
```
bash
./run-vad-remove-slience.sh
./run-ten-vad-remove-slience.sh
```
## VAD + Non-streaming Dolphin CTC for speech recognition
...
...
java-api-examples/TenVadRemoveSilence.java
0 → 100644
查看文件 @
b376ad3
// Copyright 2025 Xiaomi Corporation
// This file shows how to use a ten-vad model to remove silences from
// a wave file.
import
com.k2fsa.sherpa.onnx.*
;
import
java.util.ArrayList
;
import
java.util.Arrays
;
public
class
TenVadRemoveSilence
{
public
static
void
main
(
String
[]
args
)
{
// please download ./ten-vad.onnx from
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
String
model
=
"./ten-vad.onnx"
;
TenVadModelConfig
tenVad
=
TenVadModelConfig
.
builder
()
.
setModel
(
model
)
.
setThreshold
(
0.5f
)
.
setMinSilenceDuration
(
0.25f
)
.
setMinSpeechDuration
(
0.5f
)
.
setWindowSize
(
256
)
.
setMaxSpeechDuration
(
5.0f
)
.
build
();
VadModelConfig
config
=
VadModelConfig
.
builder
()
.
setTenVadModelConfig
(
tenVad
)
.
setSampleRate
(
16000
)
.
setNumThreads
(
1
)
.
setDebug
(
true
)
.
setProvider
(
"cpu"
)
.
build
();
Vad
vad
=
new
Vad
(
config
);
// You can download the test file from
// https://github.com/k2-fsa/sherpa-onnx/releases/tag/asr-models
String
testWaveFilename
=
"./lei-jun-test.wav"
;
WaveReader
reader
=
new
WaveReader
(
testWaveFilename
);
int
numSamples
=
reader
.
getSamples
().
length
;
int
windowSize
=
tenVad
.
getWindowSize
();
int
numIter
=
numSamples
/
windowSize
;
ArrayList
<
float
[]>
segments
=
new
ArrayList
<
float
[]>();
for
(
int
i
=
0
;
i
!=
numIter
;
++
i
)
{
int
start
=
i
*
windowSize
;
int
end
=
start
+
windowSize
;
float
[]
samples
=
Arrays
.
copyOfRange
(
reader
.
getSamples
(),
start
,
end
);
vad
.
acceptWaveform
(
samples
);
if
(
vad
.
isSpeechDetected
())
{
while
(!
vad
.
empty
())
{
// if you want to get the starting time of this segment, you can use
/* float startTime = vad.front().getStart() / 16000.0f; */
segments
.
add
(
vad
.
front
().
getSamples
());
vad
.
pop
();
}
}
}
vad
.
flush
();
while
(!
vad
.
empty
())
{
// if you want to get the starting time of this segment, you can use
/* float startTime = vad.front().getStart() / 16000.0f; */
segments
.
add
(
vad
.
front
().
getSamples
());
vad
.
pop
();
}
// get total number of samples
int
n
=
0
;
for
(
float
[]
s
:
segments
)
{
n
+=
s
.
length
;
}
float
[]
allSamples
=
new
float
[
n
];
int
i
=
0
;
for
(
float
[]
s
:
segments
)
{
System
.
arraycopy
(
s
,
0
,
allSamples
,
i
,
s
.
length
);
i
+=
s
.
length
;
}
String
outFilename
=
"lei-jun-test-no-silence.wav"
;
WaveWriter
.
write
(
outFilename
,
allSamples
,
16000
);
System
.
out
.
printf
(
"Saved to %s\n"
,
outFilename
);
vad
.
release
();
}
}
...
...
java-api-examples/run-ten-vad-remove-silence.sh
0 → 100755
查看文件 @
b376ad3
#!/usr/bin/env bash
set
-ex
if
[[
! -f ../build/lib/libsherpa-onnx-jni.dylib
&&
! -f ../build/lib/libsherpa-onnx-jni.so
]]
;
then
mkdir -p ../build
pushd
../build
cmake
\
-DSHERPA_ONNX_ENABLE_PYTHON
=
OFF
\
-DSHERPA_ONNX_ENABLE_TESTS
=
OFF
\
-DSHERPA_ONNX_ENABLE_CHECK
=
OFF
\
-DBUILD_SHARED_LIBS
=
ON
\
-DSHERPA_ONNX_ENABLE_PORTAUDIO
=
OFF
\
-DSHERPA_ONNX_ENABLE_JNI
=
ON
\
..
make -j4
ls -lh lib
popd
fi
if
[
! -f ../sherpa-onnx/java-api/build/sherpa-onnx.jar
]
;
then
pushd
../sherpa-onnx/java-api
make
popd
fi
if
[
! -f ./ten-vad.onnx
]
;
then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
fi
if
[
! -f ./lei-jun-test.wav
]
;
then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/lei-jun-test.wav
fi
java
\
-Djava.library.path
=
$PWD
/../build/lib
\
-cp ../sherpa-onnx/java-api/build/sherpa-onnx.jar
\
./TenVadRemoveSilence.java
...
...
java-api-examples/run-vad-remove-s
li
ence.sh → java-api-examples/run-vad-remove-s
il
ence.sh
查看文件 @
b376ad3
python-api-examples/simulate-streaming-sense-voice-microphone.py
查看文件 @
b376ad3
...
...
@@ -10,7 +10,7 @@ from a microphone.
Usage:
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
./python-api-examples/simulate-streaming-sense-voice-microphone.py
\
--silero-vad-model=./silero_vad.onnx
\
...
...
python-api-examples/speaker-identification-with-vad-dynamic.py
查看文件 @
b376ad3
...
...
@@ -18,12 +18,12 @@ Note that `zh` means Chinese, while `en` means English.
(2) Download the VAD model
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
(3) Run this script
...
...
python-api-examples/speaker-identification-with-vad-non-streaming-asr-alsa.py
查看文件 @
b376ad3
...
...
@@ -40,12 +40,12 @@ Note that `zh` means Chinese, while `en` means English.
(3) Download the VAD model
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
(4) Please refer to ./generate-subtitles.py
to download a non-streaming ASR model.
...
...
python-api-examples/speaker-identification-with-vad-non-streaming-asr.py
查看文件 @
b376ad3
...
...
@@ -38,12 +38,12 @@ Note that `zh` means Chinese, while `en` means English.
(3) Download the VAD model
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
(4) Please refer to ./generate-subtitles.py
to download a non-streaming ASR model.
...
...
python-api-examples/speaker-identification-with-vad.py
查看文件 @
b376ad3
...
...
@@ -36,12 +36,12 @@ Note that `zh` means Chinese, while `en` means English.
(3) Download the VAD model
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
(4) Run this script
...
...
python-api-examples/vad-alsa.py
查看文件 @
b376ad3
...
...
@@ -55,7 +55,7 @@ def main():
if
not
Path
(
args
.
silero_vad_model
)
.
is_file
():
raise
RuntimeError
(
f
"{args.silero_vad_model} does not exist. Please download it from "
"https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx"
"https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx"
)
device_name
=
args
.
device_name
...
...
python-api-examples/vad-microphone.py
查看文件 @
b376ad3
...
...
@@ -38,7 +38,7 @@ def main():
if
not
Path
(
args
.
silero_vad_model
)
.
is_file
():
raise
RuntimeError
(
f
"{args.silero_vad_model} does not exist. Please download it from "
"https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx"
"https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx"
)
mic_sample_rate
=
16000
...
...
python-api-examples/vad-remove-non-speech-segments-alsa.py
查看文件 @
b376ad3
...
...
@@ -14,12 +14,12 @@ python3 ./vad-remove-non-speech-segments-alsa.py \
--silero-vad-model silero_vad.onnx
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
"""
import
argparse
...
...
python-api-examples/vad-remove-non-speech-segments-from-file.py
查看文件 @
b376ad3
...
...
@@ -13,12 +13,11 @@ python3 ./vad-remove-non-speech-segments-from-file.py \
output.wav
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/snakers4/silero-vad/raw/master/src/silero_vad/data/silero_vad.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
"""
import
argparse
...
...
python-api-examples/vad-remove-non-speech-segments.py
查看文件 @
b376ad3
...
...
@@ -11,12 +11,12 @@ python3 ./vad-remove-non-speech-segments.py \
--silero-vad-model silero_vad.onnx
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
"""
import
argparse
...
...
python-api-examples/vad-with-non-streaming-asr.py
查看文件 @
b376ad3
...
...
@@ -70,12 +70,13 @@ to install sherpa-onnx and to download non-streaming pre-trained models
used in this file.
Please visit
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
to download silero_vad.onnx
For instance,
wget https://github.com/snakers4/silero-vad/raw/master/src/silero_vad/data/silero_vad.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
"""
import
argparse
import
sys
...
...
scripts/apk/build-apk-vad.sh
查看文件 @
b376ad3
...
...
@@ -32,11 +32,12 @@ log "====================x86===================="
mkdir -p apks
log
"https://github.com/snakers4/silero-vad/raw/master/src/silero_vad/data/silero_vad.onnx"
log
"https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx"
# Download the model
pushd
./android/SherpaOnnxVad/app/src/main/assets/
wget -c https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget -c https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
popd
for
arch
in
arm64-v8a armeabi-v7a x86_64 x86;
do
...
...
@@ -67,4 +68,47 @@ done
rm -rf ./android/SherpaOnnxVad/app/src/main/assets/
*
.onnx
# Now for ten-vad
git checkout .
pushd
android/SherpaOnnxVad/app/src/main/java/com/k2fsa/sherpa/onnx
sed -i.bak s/
"type = 0/type = 1/"
./MainActivity.kt
git diff
popd
log
"https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx"
# Download the model
pushd
./android/SherpaOnnxVad/app/src/main/assets/
wget -c https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
popd
for
arch
in
arm64-v8a armeabi-v7a x86_64 x86;
do
log
"------------------------------------------------------------"
log
"build apk for
$arch
"
log
"------------------------------------------------------------"
src_arch
=
$arch
if
[
$arch
==
"armeabi-v7a"
]
;
then
src_arch
=
armv7-eabi
elif
[
$arch
==
"x86_64"
]
;
then
src_arch
=
x86-64
fi
ls -lh ./build-android-
$src_arch
/install/lib/
*
.so
cp -v ./build-android-
$src_arch
/install/lib/
*
.so ./android/SherpaOnnxVad/app/src/main/jniLibs/
$arch
/
pushd
./android/SherpaOnnxVad
sed -i.bak s/2048/9012/g ./gradle.properties
git diff ./gradle.properties
./gradlew assembleRelease
popd
mv android/SherpaOnnxVad/app/build/outputs/apk/release/app-release-unsigned.apk ./apks/sherpa-onnx-
${
SHERPA_ONNX_VERSION
}
-
$arch
-ten_vad.apk
ls -lh apks
rm -v ./android/SherpaOnnxVad/app/src/main/jniLibs/
$arch
/
*
.so
done
rm -rf ./android/SherpaOnnxVad/app/src/main/assets/
*
.onnx
ls -lh apks/
...
...
sherpa-onnx/csrc/keyword-spotter-transducer-impl.h
查看文件 @
b376ad3
...
...
@@ -201,11 +201,11 @@ class KeywordSpotterTransducerImpl : public KeywordSpotterImpl {
int32_t
num_trailing_blanks
=
r
.
num_trailing_blanks
;
// assume subsampling_factor is 4
// assume frameshift is 0.01 second
float
trailing_s
li
ence
=
num_trailing_blanks
*
4
*
0
.
01
;
float
trailing_s
il
ence
=
num_trailing_blanks
*
4
*
0
.
01
;
// it resets automatically after detecting 1.5 seconds of silence
float
threshold
=
1
.
5
;
if
(
trailing_s
li
ence
>
threshold
)
{
if
(
trailing_s
il
ence
>
threshold
)
{
Reset
(
s
);
}
}
...
...
sherpa-onnx/csrc/sherpa-onnx-vad-alsa-offline-asr.cc
查看文件 @
b376ad3
...
...
@@ -29,10 +29,10 @@ This program shows how to use a streaming VAD with non-streaming ASR in
sherpa-onnx.
Please download silero_vad.onnx from
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
For instance, use
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
Please refer to ./sherpa-onnx-microphone-offline.cc
to download models for offline ASR.
...
...
sherpa-onnx/csrc/sherpa-onnx-vad-alsa.cc
查看文件 @
b376ad3
...
...
@@ -30,10 +30,10 @@ This program shows how to use VAD in sherpa-onnx.
device_name
Please download silero_vad.onnx from
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
For instance, use
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
The device name specifies which microphone to use in case there are several
on your system. You can use
...
...
sherpa-onnx/csrc/sherpa-onnx-vad-microphone-offline-asr.cc
查看文件 @
b376ad3
...
...
@@ -45,10 +45,10 @@ This program shows how to use a streaming VAD with non-streaming ASR in
sherpa-onnx.
Please download silero_vad.onnx from
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
For instance, use
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
Please refer to ./sherpa-onnx-microphone-offline.cc
to download models for offline ASR.
...
...
sherpa-onnx/csrc/sherpa-onnx-vad-microphone.cc
查看文件 @
b376ad3
...
...
@@ -49,10 +49,10 @@ This program shows how to use VAD in sherpa-onnx.
--vad-num-threads=1
Please download silero_vad.onnx from
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
For instance, use
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
)usage"
;
sherpa_onnx
::
ParseOptions
po
(
kUsageMessage
);
...
...
sherpa-onnx/csrc/sherpa-onnx-vad.cc
查看文件 @
b376ad3
...
...
@@ -23,10 +23,10 @@ to remove silences from a file.
/path/to/output.wav
Please download silero_vad.onnx from
https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
For instance, use
wget https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
wget https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
input.wav should be 16kHz.
)usage"
;
...
...
sherpa-onnx/java-api/Makefile
查看文件 @
b376ad3
...
...
@@ -74,6 +74,7 @@ java_files += SpeakerEmbeddingExtractorConfig.java
java_files
+=
SpeakerEmbeddingExtractor.java
java_files
+=
SpeakerEmbeddingManager.java
java_files
+=
TenVadModelConfig.java
java_files
+=
SileroVadModelConfig.java
java_files
+=
VadModelConfig.java
java_files
+=
SpeechSegment.java
...
...
sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/TenVadModelConfig.java
0 → 100644
查看文件 @
b376ad3
// Copyright 2025 Xiaomi Corporation
package
com
.
k2fsa
.
sherpa
.
onnx
;
public
class
TenVadModelConfig
{
private
final
String
model
;
private
final
float
threshold
;
private
final
float
minSilenceDuration
;
private
final
float
minSpeechDuration
;
private
final
int
windowSize
;
private
final
float
maxSpeechDuration
;
private
TenVadModelConfig
(
Builder
builder
)
{
this
.
model
=
builder
.
model
;
this
.
threshold
=
builder
.
threshold
;
this
.
minSilenceDuration
=
builder
.
minSilenceDuration
;
this
.
minSpeechDuration
=
builder
.
minSpeechDuration
;
this
.
windowSize
=
builder
.
windowSize
;
this
.
maxSpeechDuration
=
builder
.
maxSpeechDuration
;
}
public
static
Builder
builder
()
{
return
new
Builder
();
}
public
String
getModel
()
{
return
model
;
}
public
float
getThreshold
()
{
return
threshold
;
}
public
float
getMinSilenceDuration
()
{
return
minSilenceDuration
;
}
public
float
getMinSpeechDuration
()
{
return
minSpeechDuration
;
}
public
int
getWindowSize
()
{
return
windowSize
;
}
public
float
getMaxSpeechDuration
()
{
return
maxSpeechDuration
;
}
public
static
class
Builder
{
private
String
model
=
""
;
private
float
threshold
=
0.5f
;
private
float
minSilenceDuration
=
0.25f
;
private
float
minSpeechDuration
=
0.25f
;
private
int
windowSize
=
256
;
private
float
maxSpeechDuration
=
5.0f
;
public
TenVadModelConfig
build
()
{
return
new
TenVadModelConfig
(
this
);
}
public
Builder
setModel
(
String
model
)
{
this
.
model
=
model
;
return
this
;
}
public
Builder
setThreshold
(
float
threshold
)
{
this
.
threshold
=
threshold
;
return
this
;
}
public
Builder
setMinSilenceDuration
(
float
minSilenceDuration
)
{
this
.
minSilenceDuration
=
minSilenceDuration
;
return
this
;
}
public
Builder
setMinSpeechDuration
(
float
minSpeechDuration
)
{
this
.
minSpeechDuration
=
minSpeechDuration
;
return
this
;
}
public
Builder
setWindowSize
(
int
windowSize
)
{
this
.
windowSize
=
windowSize
;
return
this
;
}
public
Builder
setMaxSpeechDuration
(
float
maxSpeechDuration
)
{
this
.
maxSpeechDuration
=
maxSpeechDuration
;
return
this
;
}
}
}
...
...
sherpa-onnx/java-api/src/com/k2fsa/sherpa/onnx/VadModelConfig.java
查看文件 @
b376ad3
...
...
@@ -4,6 +4,7 @@ package com.k2fsa.sherpa.onnx;
public
class
VadModelConfig
{
private
final
SileroVadModelConfig
sileroVadModelConfig
;
private
final
TenVadModelConfig
tenVadModelConfig
;
private
final
int
sampleRate
;
private
final
int
numThreads
;
private
final
boolean
debug
;
...
...
@@ -11,6 +12,7 @@ public class VadModelConfig {
private
VadModelConfig
(
Builder
builder
)
{
this
.
sileroVadModelConfig
=
builder
.
sileroVadModelConfig
;
this
.
tenVadModelConfig
=
builder
.
tenVadModelConfig
;
this
.
sampleRate
=
builder
.
sampleRate
;
this
.
numThreads
=
builder
.
numThreads
;
this
.
debug
=
builder
.
debug
;
...
...
@@ -25,6 +27,10 @@ public class VadModelConfig {
return
sileroVadModelConfig
;
}
public
TenVadModelConfig
getTenVadModelConfig
()
{
return
tenVadModelConfig
;
}
public
int
getSampleRate
()
{
return
sampleRate
;
}
...
...
@@ -43,6 +49,7 @@ public class VadModelConfig {
public
static
class
Builder
{
private
SileroVadModelConfig
sileroVadModelConfig
=
new
SileroVadModelConfig
.
Builder
().
build
();
private
TenVadModelConfig
tenVadModelConfig
=
new
TenVadModelConfig
.
Builder
().
build
();
private
int
sampleRate
=
16000
;
private
int
numThreads
=
1
;
private
boolean
debug
=
true
;
...
...
@@ -57,6 +64,11 @@ public class VadModelConfig {
return
this
;
}
public
Builder
setTenVadModelConfig
(
TenVadModelConfig
tenVadModelConfig
)
{
this
.
tenVadModelConfig
=
tenVadModelConfig
;
return
this
;
}
public
Builder
setSampleRate
(
int
sampleRate
)
{
this
.
sampleRate
=
sampleRate
;
return
this
;
...
...
sherpa-onnx/jni/voice-activity-detector.cc
查看文件 @
b376ad3
...
...
@@ -44,6 +44,33 @@ static VadModelConfig GetVadModelConfig(JNIEnv *env, jobject config) {
ans
.
silero_vad
.
max_speech_duration
=
env
->
GetFloatField
(
silero_vad_config
,
fid
);
// ten-vad
fid
=
env
->
GetFieldID
(
cls
,
"tenVadModelConfig"
,
"Lcom/k2fsa/sherpa/onnx/TenVadModelConfig;"
);
jobject
ten_vad_config
=
env
->
GetObjectField
(
config
,
fid
);
jclass
ten_vad_config_cls
=
env
->
GetObjectClass
(
ten_vad_config
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"model"
,
"Ljava/lang/String;"
);
s
=
(
jstring
)
env
->
GetObjectField
(
ten_vad_config
,
fid
);
p
=
env
->
GetStringUTFChars
(
s
,
nullptr
);
ans
.
ten_vad
.
model
=
p
;
env
->
ReleaseStringUTFChars
(
s
,
p
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"threshold"
,
"F"
);
ans
.
ten_vad
.
threshold
=
env
->
GetFloatField
(
ten_vad_config
,
fid
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"minSilenceDuration"
,
"F"
);
ans
.
ten_vad
.
min_silence_duration
=
env
->
GetFloatField
(
ten_vad_config
,
fid
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"minSpeechDuration"
,
"F"
);
ans
.
ten_vad
.
min_speech_duration
=
env
->
GetFloatField
(
ten_vad_config
,
fid
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"windowSize"
,
"I"
);
ans
.
ten_vad
.
window_size
=
env
->
GetIntField
(
ten_vad_config
,
fid
);
fid
=
env
->
GetFieldID
(
ten_vad_config_cls
,
"maxSpeechDuration"
,
"F"
);
ans
.
ten_vad
.
max_speech_duration
=
env
->
GetFloatField
(
ten_vad_config
,
fid
);
fid
=
env
->
GetFieldID
(
cls
,
"sampleRate"
,
"I"
);
ans
.
sample_rate
=
env
->
GetIntField
(
config
,
fid
);
...
...
@@ -179,8 +206,9 @@ JNIEXPORT bool JNICALL Java_com_k2fsa_sherpa_onnx_Vad_isSpeechDetected(
}
SHERPA_ONNX_EXTERN_C
JNIEXPORT
void
JNICALL
Java_com_k2fsa_sherpa_onnx_Vad_reset
(
JNIEnv
*
env
,
jobject
/*obj*/
,
jlong
ptr
)
{
JNIEXPORT
void
JNICALL
Java_com_k2fsa_sherpa_onnx_Vad_reset
(
JNIEnv
*
env
,
jobject
/*obj*/
,
jlong
ptr
)
{
SafeJNI
(
env
,
"Vad_reset"
,
[
&
]
{
if
(
!
ValidatePointer
(
env
,
ptr
,
"Vad_reset"
,
"VoiceActivityDetector pointer is null."
))
{
...
...
sherpa-onnx/kotlin-api/Vad.kt
查看文件 @
b376ad3
...
...
@@ -12,8 +12,18 @@ data class SileroVadModelConfig(
var maxSpeechDuration: Float = 5.0F,
)
data class TenVadModelConfig(
var model: String = "",
var threshold: Float = 0.5F,
var minSilenceDuration: Float = 0.25F,
var minSpeechDuration: Float = 0.25F,
var windowSize: Int = 256,
var maxSpeechDuration: Float = 5.0F,
)
data class VadModelConfig(
var sileroVadModelConfig: SileroVadModelConfig = SileroVadModelConfig(),
var tenVadModelConfig: TenVadModelConfig = TenVadModelConfig(),
var sampleRate: Int = 16000,
var numThreads: Int = 1,
var provider: String = "cpu",
...
...
@@ -91,10 +101,14 @@ class Vad(
}
// Please visit
// https://github.com/
snakers4/silero-vad/raw/master/src/silero_vad/data
/silero_vad.onnx
// https://github.com/
k2-fsa/sherpa-onnx/releases/download/asr-models
/silero_vad.onnx
// to download silero_vad.onnx
// and put it inside the assets/
// directory
//
// For ten-vad, please use
// https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/ten-vad.onnx
//
fun getVadModelConfig(type: Int): VadModelConfig? {
when (type) {
0 -> {
...
...
@@ -111,6 +125,21 @@ fun getVadModelConfig(type: Int): VadModelConfig? {
provider = "cpu",
)
}
1 -> {
return VadModelConfig(
tenVadModelConfig = TenVadModelConfig(
model = "ten-vad.onnx",
threshold = 0.5F,
minSilenceDuration = 0.25F,
minSpeechDuration = 0.25F,
windowSize = 256,
),
sampleRate = 16000,
numThreads = 1,
provider = "cpu",
)
}
}
return null
}
...
...
请
注册
或
登录
后发表评论