xuning / sherpaonnx

转到一个项目
Toggle navigation Toggle navigation pinning
Fangjun Kuang
Committed by GitHub
0aacf02d 1 parent 623cdc9e

Add C++ runtime for vocos (#2014)

内嵌 并排对比
正在显示 62 个修改的文件 包含 558 行增加 和 162 行删除
... ... @@ -121,19 +121,19 @@ rm -rf kokoro-en-v0_19
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-icefall-en_US-ljspeech.tar.bz2
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test_tts_non_streaming_matcha_icefall_en.js
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm -rf matcha-icefall-en_US-ljspeech
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-icefall-zh-baker.tar.bz2
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test_tts_non_streaming_matcha_icefall_zh.js
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm -rf matcha-icefall-zh-baker
ls -lh *.wav
... ...
... ... @@ -42,12 +42,13 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test-offline-tts-matcha-zh.js
rm -rf matcha-icefall-zh-baker
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
echo "---"
... ... @@ -55,12 +56,12 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test-offline-tts-matcha-en.js
rm -rf matcha-icefall-en_US-ljspeech
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
echo "---"
... ...
... ... @@ -50,11 +50,12 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-tts-fa_en-musa.tar.bz2
rm matcha-tts-fa_en-musa.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
$EXE \
--matcha-acoustic-model=./matcha-tts-fa_en-musa/model.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-tts-fa_en-musa/tokens.txt \
--matcha-data-dir=./matcha-tts-fa_en-musa/espeak-ng-data \
--output-filename=./tts/test-matcha-fa-en-musa.wav \
... ... @@ -62,7 +63,7 @@ $EXE \
"How are you doing today? این یک نمونه ی تست فارسی است. This is a test."
rm -rf matcha-tts-fa_en-musa
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
ls -lh tts/*.wav
log "------------------------------------------------------------"
... ... @@ -72,11 +73,12 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
$EXE \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--num-threads=2 \
... ... @@ -84,7 +86,7 @@ $EXE \
--debug=1 \
"Today as always, men fall into two groups: slaves and free men. Whoever does not have two-thirds of his day for himself, is a slave, whatever he may be: a statesman, a businessman, an official, or a scholar."
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm -rf matcha-icefall-en_US-ljspeech
ls -lh tts/*.wav
... ... @@ -95,11 +97,11 @@ curl -O -SL https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
$EXE \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--matcha-dict-dir=./matcha-icefall-zh-baker/dict \
... ... @@ -110,7 +112,7 @@ $EXE \
$EXE \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--matcha-dict-dir=./matcha-icefall-zh-baker/dict \
... ... @@ -119,7 +121,7 @@ $EXE \
--output-filename=./tts/matcha-baker-zh-2.wav \
"当夜幕降临，星光点点，伴随着微风拂面，我在静谧中感受着时光的流转，思念如涟漪荡漾，梦境如画卷展开，我与自然融为一体，沉静在这片宁静的美丽之中，感受着生命的奇迹与温柔。"
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm -rf matcha-icefall-zh-baker
log "------------------------------------------------------------"
... ...
... ... @@ -320,18 +320,18 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts.py \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--output-filename=./tts/test-matcha-ljspeech-en.wav \
--num-threads=2 \
"Today as always, men fall into two groups: slaves and free men. Whoever does not have two-thirds of his day for himself, is a slave, whatever he may be: a statesman, a businessman, an official, or a scholar."
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm -rf matcha-icefall-en_US-ljspeech
log "matcha-baker-zh test"
... ... @@ -340,11 +340,11 @@ curl -O -SL https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts.py \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--tts-rule-fsts=./matcha-icefall-zh-baker/phone.fst,./matcha-icefall-zh-baker/date.fst,./matcha-icefall-zh-baker/number.fst \
... ... @@ -353,7 +353,7 @@ python3 ./python-api-examples/offline-tts.py \
"某某银行的副行长和一些行政领导表示，他们去过长江和长白山; 经济不断增长。2024年12月31号，拨打110或者18920240511。123456块钱。"
rm -rf matcha-icefall-zh-baker
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
log "vits-ljs test"
... ...
... ... @@ -228,7 +228,7 @@ jobs:
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
export LD_LIBRARY_PATH=$PWD/build/install/lib:$LD_LIBRARY_PATH
export DYLD_LIBRARY_PATH=$PWD/build/install/lib:$DYLD_LIBRARY_PATH
... ... @@ -237,7 +237,7 @@ jobs:
rm ./matcha-tts-zh-c-api
rm -rf matcha-icefall-*
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
- name: Test Matcha TTS (en)
shell: bash
... ... @@ -252,7 +252,7 @@ jobs:
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
export LD_LIBRARY_PATH=$PWD/build/install/lib:$LD_LIBRARY_PATH
export DYLD_LIBRARY_PATH=$PWD/build/install/lib:$DYLD_LIBRARY_PATH
... ... @@ -261,7 +261,7 @@ jobs:
rm ./matcha-tts-en-c-api
rm -rf matcha-icefall-*
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
- uses: actions/upload-artifact@v4
with:
... ...
... ... @@ -237,7 +237,7 @@ jobs:
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
export LD_LIBRARY_PATH=$PWD/build/install/lib:$LD_LIBRARY_PATH
export DYLD_LIBRARY_PATH=$PWD/build/install/lib:$DYLD_LIBRARY_PATH
... ... @@ -245,7 +245,7 @@ jobs:
./matcha-tts-zh-cxx-api
rm -rf matcha-icefall-*
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
rm matcha-tts-zh-cxx-api
- name: Test Matcha TTS (en)
... ... @@ -262,7 +262,7 @@ jobs:
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
export LD_LIBRARY_PATH=$PWD/build/install/lib:$LD_LIBRARY_PATH
export DYLD_LIBRARY_PATH=$PWD/build/install/lib:$DYLD_LIBRARY_PATH
... ... @@ -271,7 +271,7 @@ jobs:
rm matcha-tts-en-cxx-api
rm -rf matcha-icefall-*
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
- uses: actions/upload-artifact@v4
with:
... ...
... ... @@ -265,7 +265,7 @@ jobs:
rm -rf kokoro-en-*
rm -rf matcha-icefall-*
rm hifigan_v2.onnx
rm vocos-22khz-univ.onnx
./run-non-streaming-tts-piper-en.sh
rm -rf vits-piper-*
... ...
... ... @@ -262,7 +262,7 @@ class MainActivity : AppCompatActivity() {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
// modelDir = "matcha-icefall-zh-baker"
// acousticModelName = "model-steps-3.onnx"
// vocoder = "hifigan_v2.onnx"
// vocoder = "vocos-22khz-univ.onnx"
// lexicon = "lexicon.txt"
// dictDir = "matcha-icefall-zh-baker/dict"
... ... @@ -271,7 +271,7 @@ class MainActivity : AppCompatActivity() {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
// modelDir = "matcha-icefall-en_US-ljspeech"
// acousticModelName = "model-steps-3.onnx"
// vocoder = "hifigan_v2.onnx"
// vocoder = "vocos-22khz-univ.onnx"
// dataDir = "matcha-icefall-en_US-ljspeech/espeak-ng-data"
// Example 9
... ...
... ... @@ -131,7 +131,7 @@ object TtsEngine {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
// modelDir = "matcha-icefall-zh-baker"
// acousticModelName = "model-steps-3.onnx"
// vocoder = "hifigan_v2.onnx"
// vocoder = "vocos-22khz-univ.onnx"
// lexicon = "lexicon.txt"
// dictDir = "matcha-icefall-zh-baker/dict"
// lang = "zho"
... ... @@ -141,7 +141,7 @@ object TtsEngine {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
// modelDir = "matcha-icefall-en_US-ljspeech"
// acousticModelName = "model-steps-3.onnx"
// vocoder = "hifigan_v2.onnx"
// vocoder = "vocos-22khz-univ.onnx"
// dataDir = "matcha-icefall-en_US-ljspeech/espeak-ng-data"
// lang = "eng"
... ...
... ... @@ -13,7 +13,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
./matcha-tts-en-c-api
... ... @@ -40,7 +40,7 @@ int32_t main(int32_t argc, char *argv[]) {
config.model.matcha.acoustic_model =
"./matcha-icefall-en_US-ljspeech/model-steps-3.onnx";
config.model.matcha.vocoder = "./hifigan_v2.onnx";
config.model.matcha.vocoder = "./vocos-22khz-univ.onnx";
config.model.matcha.tokens = "./matcha-icefall-en_US-ljspeech/tokens.txt";
... ...
... ... @@ -13,7 +13,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
./matcha-tts-zh-c-api
... ... @@ -39,7 +39,7 @@ int32_t main(int32_t argc, char *argv[]) {
memset(&config, 0, sizeof(config));
config.model.matcha.acoustic_model =
"./matcha-icefall-zh-baker/model-steps-3.onnx";
config.model.matcha.vocoder = "./hifigan_v2.onnx";
config.model.matcha.vocoder = "./vocos-22khz-univ.onnx";
config.model.matcha.lexicon = "./matcha-icefall-zh-baker/lexicon.txt";
config.model.matcha.tokens = "./matcha-icefall-zh-baker/tokens.txt";
config.model.matcha.dict_dir = "./matcha-icefall-zh-baker/dict";
... ...
... ... @@ -13,7 +13,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
./matcha-tts-en-cxx-api
... ... @@ -39,7 +39,7 @@ int32_t main(int32_t argc, char *argv[]) {
config.model.matcha.acoustic_model =
"./matcha-icefall-en_US-ljspeech/model-steps-3.onnx";
config.model.matcha.vocoder = "./hifigan_v2.onnx";
config.model.matcha.vocoder = "./vocos-22khz-univ.onnx";
config.model.matcha.tokens = "./matcha-icefall-en_US-ljspeech/tokens.txt";
... ...
... ... @@ -13,7 +13,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
./matcha-tts-zh-cxx-api
... ... @@ -37,7 +37,7 @@ int32_t main(int32_t argc, char *argv[]) {
OfflineTtsConfig config;
config.model.matcha.acoustic_model =
"./matcha-icefall-zh-baker/model-steps-3.onnx";
config.model.matcha.vocoder = "./hifigan_v2.onnx";
config.model.matcha.vocoder = "./vocos-22khz-univ.onnx";
config.model.matcha.lexicon = "./matcha-icefall-zh-baker/lexicon.txt";
config.model.matcha.tokens = "./matcha-icefall-zh-baker/tokens.txt";
config.model.matcha.dict_dir = "./matcha-icefall-zh-baker/dict";
... ...
... ... @@ -14,14 +14,14 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dart run \
./bin/matcha-en.dart \
--acoustic-model ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--vocoder ./hifigan_v2.onnx \
--vocoder ./vocos-22khz-univ.onnx \
--tokens ./matcha-icefall-en_US-ljspeech/tokens.txt \
--data-dir ./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--sid 0 \
... ...
... ... @@ -13,14 +13,14 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dart run \
./bin/matcha-zh.dart \
--acoustic-model ./matcha-icefall-zh-baker/model-steps-3.onnx \
--vocoder ./hifigan_v2.onnx \
--vocoder ./vocos-22khz-univ.onnx \
--lexicon ./matcha-icefall-zh-baker/lexicon.txt \
--tokens ./matcha-icefall-zh-baker/tokens.txt \
--dict-dir ./matcha-icefall-zh-baker/dict \
... ... @@ -33,7 +33,7 @@ dart run \
dart run \
./bin/matcha-zh.dart \
--acoustic-model ./matcha-icefall-zh-baker/model-steps-3.onnx \
--vocoder ./hifigan_v2.onnx \
--vocoder ./vocos-22khz-univ.onnx \
--lexicon ./matcha-icefall-zh-baker/lexicon.txt \
--tokens ./matcha-icefall-zh-baker/tokens.txt \
--dict-dir ./matcha-icefall-zh-baker/dict \
... ...
... ... @@ -92,11 +92,11 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--dict-dir=./matcha-icefall-zh-baker/dict \
... ... @@ -111,11 +111,11 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
... ...
... ... @@ -12,13 +12,13 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dotnet run \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
... ...
... ... @@ -10,14 +10,14 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dotnet run \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--dict-dir=./matcha-icefall-zh-baker/dict \
... ...
... ... @@ -85,11 +85,11 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--dict-dir=./matcha-icefall-zh-baker/dict \
... ... @@ -104,11 +104,11 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
dotnet run \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
... ...
... ... @@ -12,13 +12,13 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dotnet run \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
... ...
... ... @@ -10,14 +10,14 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
dotnet run \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--tokens=./matcha-icefall-zh-baker/tokens.txt \
--dict-dir=./matcha-icefall-zh-baker/dict \
... ...
... ... @@ -12,8 +12,8 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
go mod tidy
... ... @@ -21,7 +21,7 @@ go build
./non-streaming-tts \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--debug=1 \
... ...
... ... @@ -11,8 +11,8 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
go mod tidy
... ... @@ -20,7 +20,7 @@ go build
./non-streaming-tts \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--matcha-dict-dir=./matcha-icefall-zh-baker/dict \
... ...
... ... @@ -159,7 +159,7 @@ function initTts(context: Context): OfflineTts {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
// modelDir = 'matcha-icefall-zh-baker';
// acousticModelName = 'model-steps-3.onnx';
// vocoder = 'hifigan_v2.onnx';
// vocoder = 'vocos-22khz-univ.onnx';
// lexicon = 'lexicon.txt';
// dictDir = 'dict';
// ruleFsts = `date.fst,phone.fst,number.fst`;
... ... @@ -169,7 +169,7 @@ function initTts(context: Context): OfflineTts {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
// modelDir = 'matcha-icefall-en_US-ljspeech';
// acousticModelName = 'model-steps-3.onnx';
// vocoder = 'hifigan_v2.onnx';
// vocoder = 'vocos-22khz-univ.onnx';
// dataDir = 'espeak-ng-data';
// Example 10
... ...
... ... @@ -131,7 +131,7 @@ func getTtsFor_matcha_icefall_zh_baker() -> SherpaOnnxOfflineTtsWrapper {
// please see https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
let acousticModel = getResource("model-steps-3", "onnx")
let vocoder = getResource("hifigan_v2", "onnx")
let vocoder = getResource("vocos-22khz-univ", "onnx")
let tokens = getResource("tokens", "txt")
let lexicon = getResource("lexicon", "txt")
... ...
... ... @@ -10,7 +10,7 @@ public class NonStreamingTtsMatchaEn {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-en-us-ljspeech-american-english-1-female-speaker
// to download model files
String acousticModel = "./matcha-icefall-en_US-ljspeech/model-steps-3.onnx";
String vocoder = "./hifigan_v2.onnx";
String vocoder = "./vocos-22khz-univ.onnx";
String tokens = "./matcha-icefall-en_US-ljspeech/tokens.txt";
String dataDir = "./matcha-icefall-en_US-ljspeech/espeak-ng-data";
String text =
... ...
... ... @@ -10,7 +10,7 @@ public class NonStreamingTtsMatchaZh {
// https://k2-fsa.github.io/sherpa/onnx/tts/pretrained_models/matcha.html#matcha-icefall-zh-baker-chinese-1-female-speaker
// to download model files
String acousticModel = "./matcha-icefall-zh-baker/model-steps-3.onnx";
String vocoder = "./hifigan_v2.onnx";
String vocoder = "./vocos-22khz-univ.onnx";
String tokens = "./matcha-icefall-zh-baker/tokens.txt";
String lexicon = "./matcha-icefall-zh-baker/lexicon.txt";
String dictDir = "./matcha-icefall-zh-baker/dict";
... ...
... ... @@ -35,8 +35,8 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
java \
... ...
... ... @@ -34,8 +34,8 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
java \
... ...
... ... @@ -111,8 +111,8 @@ function testTts() {
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
if [ ! -f ./kokoro-multi-lang-v1_0/model.onnx ]; then
... ...
... ... @@ -58,7 +58,7 @@ fun testMatcha() {
model=OfflineTtsModelConfig(
matcha=OfflineTtsMatchaModelConfig(
acousticModel="./matcha-icefall-zh-baker/model-steps-3.onnx",
vocoder="./hifigan_v2.onnx",
vocoder="./vocos-22khz-univ.onnx",
tokens="./matcha-icefall-zh-baker/tokens.txt",
lexicon="./matcha-icefall-zh-baker/lexicon.txt",
dictDir="./matcha-icefall-zh-baker/dict",
... ...
... ... @@ -513,9 +513,9 @@ void CNonStreamingTextToSpeechDlg::Init() {
"(c) Switch to the directory matcha-icefall-zh-baker\r\n"
"(d) Rename model-steps-3.onnx to model.onnx\r\n"
"(e) Download a vocoder model from \r\n"
" https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx\r\n"
"(f) Rename hifigan_v2.onnx to hifigan.onnx\r\n"
"(g) Remember to put hifigan.onnx in the directory matcha-icefall-zh-baker\r\n"
" https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx\r\n"
"(f) Rename vocos-22khz-univ.onnx to vocos.onnx\r\n"
"(g) Remember to put vocos.onnx in the directory matcha-icefall-zh-baker\r\n"
"(h) Copy the current exe to the directory matcha-icefall-zh-baker\r\n"
"(i) Done! You can now run the exe in the directory matcha-icefall-zh-baker\r\n";
... ... @@ -540,10 +540,16 @@ void CNonStreamingTextToSpeechDlg::Init() {
config.model.kokoro.dict_dir = "./dict";
config.model.kokoro.lexicon = "./lexicon-us-en.txt,./lexicon-zh.txt";
}
} else if (Exists("./hifigan.onnx")) {
} else if (Exists("./hifigan.onnx") || Exists("./vocos.onnx")) {
// it is a matcha tts model
config.model.matcha.acoustic_model = "./model.onnx";
config.model.matcha.vocoder = "./hifigan.onnx";
if (Exists("./hifigan.onnx")) {
config.model.matcha.vocoder = "./hifigan.onnx";
} else if (Exists("./vocos.onnx")) {
config.model.matcha.vocoder = "./vocos.onnx";
}
config.model.matcha.tokens = "./tokens.txt";
if (Exists("./espeak-ng-data/phontab")) {
... ...
... ... @@ -400,7 +400,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test_tts_non_streaming_matcha_icefall_en.js
```
... ... @@ -411,7 +411,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test_tts_non_streaming_matcha_icefall_zh.js
```
... ...
... ... @@ -9,7 +9,7 @@ function createOfflineTts() {
model: {
matcha: {
acousticModel: './matcha-icefall-en_US-ljspeech/model-steps-3.onnx',
vocoder: './hifigan_v2.onnx',
vocoder: './vocos-22khz-univ.onnx',
lexicon: './matcha-icefall-en_US-ljspeech/lexicon.txt',
tokens: './matcha-icefall-en_US-ljspeech/tokens.txt',
dataDir: './matcha-icefall-en_US-ljspeech/espeak-ng-data',
... ...
... ... @@ -9,7 +9,7 @@ function createOfflineTts() {
model: {
matcha: {
acousticModel: './matcha-icefall-zh-baker/model-steps-3.onnx',
vocoder: './hifigan_v2.onnx',
vocoder: './vocos-22khz-univ.onnx',
lexicon: './matcha-icefall-zh-baker/lexicon.txt',
tokens: './matcha-icefall-zh-baker/tokens.txt',
dictDir: './matcha-icefall-zh-baker/dict',
... ...
... ... @@ -83,7 +83,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test-offline-tts-matcha-zh.js
```
... ... @@ -101,7 +101,7 @@ wget https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/matcha-i
tar xf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
wget https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
node ./test-offline-tts-matcha-en.js
```
... ...
... ... @@ -5,7 +5,7 @@ const sherpa_onnx = require('sherpa-onnx');
function createOfflineTts() {
let offlineTtsMatchaModelConfig = {
acousticModel: './matcha-icefall-en_US-ljspeech/model-steps-3.onnx',
vocoder: './hifigan_v2.onnx',
vocoder: './vocos-22khz-univ.onnx',
lexicon: './matcha-icefall-en_US-ljspeech/lexicon.txt',
tokens: './matcha-icefall-en_US-ljspeech/tokens.txt',
dataDir: './matcha-icefall-en_US-ljspeech/espeak-ng-data',
... ...
... ... @@ -5,7 +5,7 @@ const sherpa_onnx = require('sherpa-onnx');
function createOfflineTts() {
let offlineTtsMatchaModelConfig = {
acousticModel: './matcha-icefall-zh-baker/model-steps-3.onnx',
vocoder: './hifigan_v2.onnx',
vocoder: './vocos-22khz-univ.onnx',
lexicon: './matcha-icefall-zh-baker/lexicon.txt',
tokens: './matcha-icefall-zh-baker/tokens.txt',
dictDir: './matcha-icefall-zh-baker/dict',
... ...
... ... @@ -115,7 +115,7 @@ var
Config: TSherpaOnnxOfflineTtsConfig;
begin
Config.Model.Matcha.AcousticModel := './matcha-icefall-en_US-ljspeech/model-steps-3.onnx';
Config.Model.Matcha.Vocoder := './hifigan_v2.onnx';
Config.Model.Matcha.Vocoder := './vocos-22khz-univ.onnx';
Config.Model.Matcha.Tokens := './matcha-icefall-en_US-ljspeech/tokens.txt';
Config.Model.Matcha.DataDir := './matcha-icefall-en_US-ljspeech/espeak-ng-data';
Config.Model.NumThreads := 1;
... ...
... ... @@ -21,7 +21,7 @@ var
Config: TSherpaOnnxOfflineTtsConfig;
begin
Config.Model.Matcha.AcousticModel := './matcha-icefall-en_US-ljspeech/model-steps-3.onnx';
Config.Model.Matcha.Vocoder := './hifigan_v2.onnx';
Config.Model.Matcha.Vocoder := './vocos-22khz-univ.onnx';
Config.Model.Matcha.Tokens := './matcha-icefall-en_US-ljspeech/tokens.txt';
Config.Model.Matcha.DataDir := './matcha-icefall-en_US-ljspeech/espeak-ng-data';
Config.Model.NumThreads := 1;
... ...
... ... @@ -115,7 +115,7 @@ var
Config: TSherpaOnnxOfflineTtsConfig;
begin
Config.Model.Matcha.AcousticModel := './matcha-icefall-zh-baker/model-steps-3.onnx';
Config.Model.Matcha.Vocoder := './hifigan_v2.onnx';
Config.Model.Matcha.Vocoder := './vocos-22khz-univ.onnx';
Config.Model.Matcha.Lexicon := './matcha-icefall-zh-baker/lexicon.txt';
Config.Model.Matcha.Tokens := './matcha-icefall-zh-baker/tokens.txt';
Config.Model.Matcha.DictDir := './matcha-icefall-zh-baker/dict';
... ...
... ... @@ -21,7 +21,7 @@ var
Config: TSherpaOnnxOfflineTtsConfig;
begin
Config.Model.Matcha.AcousticModel := './matcha-icefall-zh-baker/model-steps-3.onnx';
Config.Model.Matcha.Vocoder := './hifigan_v2.onnx';
Config.Model.Matcha.Vocoder := './vocos-22khz-univ.onnx';
Config.Model.Matcha.Lexicon := './matcha-icefall-zh-baker/lexicon.txt';
Config.Model.Matcha.Tokens := './matcha-icefall-zh-baker/tokens.txt';
Config.Model.Matcha.DictDir := './matcha-icefall-zh-baker/dict';
... ...
... ... @@ -33,8 +33,8 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
fpc \
... ...
... ... @@ -33,8 +33,8 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
fpc \
... ...
... ... @@ -32,8 +32,8 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
fpc \
... ...
... ... @@ -32,8 +32,8 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
fpc \
... ...
... ... @@ -59,11 +59,11 @@ curl -O -SL https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts-play.py \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--tts-rule-fsts=./matcha-icefall-zh-baker/phone.fst,./matcha-icefall-zh-baker/date.fst,./matcha-icefall-zh-baker/number.fst \
... ... @@ -77,11 +77,11 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts-play.py \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--output-filename=./test-matcha-ljspeech-en.wav \
... ...
... ... @@ -60,11 +60,11 @@ curl -O -SL https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-zh-baker.tar.bz2
rm matcha-icefall-zh-baker.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts.py \
--matcha-acoustic-model=./matcha-icefall-zh-baker/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-lexicon=./matcha-icefall-zh-baker/lexicon.txt \
--matcha-tokens=./matcha-icefall-zh-baker/tokens.txt \
--tts-rule-fsts=./matcha-icefall-zh-baker/phone.fst,./matcha-icefall-zh-baker/date.fst,./matcha-icefall-zh-baker/number.fst \
... ... @@ -78,11 +78,11 @@ curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/tts-models/m
tar xvf matcha-icefall-en_US-ljspeech.tar.bz2
rm matcha-icefall-en_US-ljspeech.tar.bz2
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
python3 ./python-api-examples/offline-tts.py \
--matcha-acoustic-model=./matcha-icefall-en_US-ljspeech/model-steps-3.onnx \
--matcha-vocoder=./hifigan_v2.onnx \
--matcha-vocoder=./vocos-22khz-univ.onnx \
--matcha-tokens=./matcha-icefall-en_US-ljspeech/tokens.txt \
--matcha-data-dir=./matcha-icefall-en_US-ljspeech/espeak-ng-data \
--output-filename=./test-matcha-ljspeech-en.wav \
... ...
... ... @@ -395,7 +395,7 @@ def get_matcha_models() -> List[TtsModel]:
s = [f"{m.model_dir}/{r}" for r in rule_fsts]
m.rule_fsts = ",".join(s)
m.dict_dir = m.model_dir + "/dict"
m.vocoder = "hifigan_v2.onnx"
m.vocoder = "vocos-22khz-univ.onnx"
english_persian_models = [
TtsModel(
... ... @@ -416,7 +416,7 @@ def get_matcha_models() -> List[TtsModel]:
]
for m in english_persian_models:
m.data_dir = f"{m.model_dir}/espeak-ng-data"
m.vocoder = "hifigan_v2.onnx"
m.vocoder = "vocos-22khz-univ.onnx"
return chinese_models + english_persian_models
... ...
... ... @@ -20,8 +20,8 @@ if [ ! -f male/tokens.txt ]; then
curl -SL --output male/tokens.txt https://huggingface.co/mah92/Musa-FA_EN-Matcha-TTS-Model/resolve/main/tokens_sherpa_with_fa.txt
fi
if [ ! -f hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
if [ ! -f .add-meta-data.done ]; then
... ... @@ -31,14 +31,14 @@ fi
python3 ./test.py \
--am ./female/model.onnx \
--vocoder ./hifigan_v2.onnx \
--vocoder ./vocos-22khz-univ.onnx \
--tokens ./female/tokens.txt \
--text "This is a test. این یک نمونه ی تست فارسی است." \
--out-wav "./female-en-fa.wav"
python3 ./test.py \
--am ./male/model.onnx \
--vocoder ./hifigan_v2.onnx \
--vocoder ./vocos-22khz-univ.onnx \
--tokens ./male/tokens.txt \
--text "This is a test. این یک نمونه ی تست فارسی است." \
--out-wav "./male-en-fa.wav"
... ...
... ... @@ -183,6 +183,8 @@ if(SHERPA_ONNX_ENABLE_TTS)
offline-tts-vits-model.cc
offline-tts.cc
piper-phonemize-lexicon.cc
vocoder.cc
vocos-vocoder.cc
)
endif()
... ...
... ... @@ -45,11 +45,21 @@ class HifiganVocoder::Impl {
Init(buf.data(), buf.size());
}
Ort::Value Run(Ort::Value mel) const {
std::vector<float> Run(Ort::Value mel) const {
auto out = sess_->Run({}, input_names_ptr_.data(), &mel, 1,
output_names_ptr_.data(), output_names_ptr_.size());
return std::move(out[0]);
std::vector<int64_t> audio_shape =
out[0].GetTensorTypeAndShapeInfo().GetShape();
int64_t total = 1;
// The output shape may be (1, 1, total) or (1, total) or (total,)
for (auto i : audio_shape) {
total *= i;
}
const float *p = out[0].GetTensorData<float>();
return {p, p + total};
}
private:
... ... @@ -88,7 +98,7 @@ HifiganVocoder::HifiganVocoder(Manager *mgr, int32_t num_threads,
HifiganVocoder::~HifiganVocoder() = default;
Ort::Value HifiganVocoder::Run(Ort::Value mel) const {
std::vector<float> HifiganVocoder::Run(Ort::Value mel) const {
return impl_->Run(std::move(mel));
}
... ...
... ... @@ -7,14 +7,16 @@
#include <memory>
#include <string>
#include <vector>
#include "onnxruntime_cxx_api.h" // NOLINT
#include "sherpa-onnx/csrc/vocoder.h"
namespace sherpa_onnx {
class HifiganVocoder {
class HifiganVocoder : public Vocoder {
public:
~HifiganVocoder();
~HifiganVocoder() override;
HifiganVocoder(int32_t num_threads, const std::string &provider,
const std::string &model);
... ... @@ -26,7 +28,7 @@ class HifiganVocoder {
/** @param mel A float32 tensor of shape (batch_size, feat_dim, num_frames).
* @return Return a float32 tensor of shape (batch_size, num_samples).
*/
Ort::Value Run(Ort::Value mel) const;
std::vector<float> Run(Ort::Value mel) const override;
private:
class Impl;
... ...
... ... @@ -13,7 +13,6 @@
#include "fst/extensions/far/far.h"
#include "kaldifst/csrc/kaldi-fst-io.h"
#include "kaldifst/csrc/text-normalizer.h"
#include "sherpa-onnx/csrc/hifigan-vocoder.h"
#include "sherpa-onnx/csrc/jieba-lexicon.h"
#include "sherpa-onnx/csrc/lexicon.h"
#include "sherpa-onnx/csrc/macros.h"
... ... @@ -25,6 +24,7 @@
#include "sherpa-onnx/csrc/onnx-utils.h"
#include "sherpa-onnx/csrc/piper-phonemize-lexicon.h"
#include "sherpa-onnx/csrc/text-utils.h"
#include "sherpa-onnx/csrc/vocoder.h"
namespace sherpa_onnx {
... ... @@ -33,9 +33,7 @@ class OfflineTtsMatchaImpl : public OfflineTtsImpl {
explicit OfflineTtsMatchaImpl(const OfflineTtsConfig &config)
: config_(config),
model_(std::make_unique<OfflineTtsMatchaModel>(config.model)),
vocoder_(std::make_unique<HifiganVocoder>(
config.model.num_threads, config.model.provider,
config.model.matcha.vocoder)) {
vocoder_(Vocoder::Create(config.model)) {
InitFrontend();
if (!config.rule_fsts.empty()) {
... ... @@ -92,9 +90,7 @@ class OfflineTtsMatchaImpl : public OfflineTtsImpl {
OfflineTtsMatchaImpl(Manager *mgr, const OfflineTtsConfig &config)
: config_(config),
model_(std::make_unique<OfflineTtsMatchaModel>(mgr, config.model)),
vocoder_(std::make_unique<HifiganVocoder>(
mgr, config.model.num_threads, config.model.provider,
config.model.matcha.vocoder)) {
vocoder_(Vocoder::Create(mgr, config.model)) {
InitFrontend(mgr);
if (!config.rule_fsts.empty()) {
... ... @@ -382,22 +378,11 @@ class OfflineTtsMatchaImpl : public OfflineTtsImpl {
memory_info, x.data(), x.size(), x_shape.data(), x_shape.size());
Ort::Value mel = model_->Run(std::move(x_tensor), sid, speed);
Ort::Value audio = vocoder_->Run(std::move(mel));
std::vector<int64_t> audio_shape =
audio.GetTensorTypeAndShapeInfo().GetShape();
int64_t total = 1;
// The output shape may be (1, 1, total) or (1, total) or (total,)
for (auto i : audio_shape) {
total *= i;
}
const float *p = audio.GetTensorData<float>();
GeneratedAudio ans;
ans.samples = vocoder_->Run(std::move(mel));
ans.sample_rate = model_->GetMetaData().sample_rate;
ans.samples = std::vector<float>(p, p + total);
float silence_scale = config_.silence_scale;
if (silence_scale != 1) {
... ... @@ -410,7 +395,7 @@ class OfflineTtsMatchaImpl : public OfflineTtsImpl {
private:
OfflineTtsConfig config_;
std::unique_ptr<OfflineTtsMatchaModel> model_;
std::unique_ptr<HifiganVocoder> vocoder_;
std::unique_ptr<Vocoder> vocoder_;
std::vector<std::unique_ptr<kaldifst::TextNormalizer>> tn_list_;
std::unique_ptr<OfflineTtsFrontend> frontend_;
};
... ...
// sherpa-onnx/csrc/vocoder.cc
//
// Copyright (c) 2025 Xiaomi Corporation
#include "sherpa-onnx/csrc/vocoder.h"
#if __ANDROID_API__ >= 9
#include "android/asset_manager.h"
#include "android/asset_manager_jni.h"
#endif
#if __OHOS__
#include "rawfile/raw_file_manager.h"
#endif
#include "sherpa-onnx/csrc/file-utils.h"
#include "sherpa-onnx/csrc/hifigan-vocoder.h"
#include "sherpa-onnx/csrc/macros.h"
#include "sherpa-onnx/csrc/onnx-utils.h"
#include "sherpa-onnx/csrc/vocos-vocoder.h"
namespace sherpa_onnx {
namespace {
enum class ModelType : std::uint8_t {
kHifigan,
kVocoos,
kUnknown,
};
} // namespace
static ModelType GetModelType(char *model_data, size_t model_data_length,
bool debug) {
Ort::Env env(ORT_LOGGING_LEVEL_ERROR);
Ort::SessionOptions sess_opts;
sess_opts.SetIntraOpNumThreads(1);
sess_opts.SetInterOpNumThreads(1);
auto sess = std::make_unique<Ort::Session>(env, model_data, model_data_length,
sess_opts);
Ort::ModelMetadata meta_data = sess->GetModelMetadata();
if (debug) {
std::ostringstream os;
PrintModelMetadata(os, meta_data);
#if __OHOS__
SHERPA_ONNX_LOGE("%{public}s", os.str().c_str());
#else
SHERPA_ONNX_LOGE("%s", os.str().c_str());
#endif
}
Ort::AllocatorWithDefaultOptions allocator;
auto model_type =
LookupCustomModelMetaData(meta_data, "model_type", allocator);
if (model_type.empty()) {
SHERPA_ONNX_LOGE(
"No model_type in the metadata!\n"
"Please make sure you are using the vocoder from "
"https://github.com/k2-fsa/sherpa-onnx/releases/tag/vocoder-models");
return ModelType::kUnknown;
}
if (model_type == "hifigan") {
return ModelType::kHifigan;
} else if (model_type == "vocos") {
return ModelType::kVocoos;
} else {
SHERPA_ONNX_LOGE("Unsupported model_type: %s", model_type.c_str());
return ModelType::kUnknown;
}
}
std::unique_ptr<Vocoder> Vocoder::Create(const OfflineTtsModelConfig &config) {
auto buffer = ReadFile(config.matcha.vocoder);
auto model_type = GetModelType(buffer.data(), buffer.size(), config.debug);
switch (model_type) {
case ModelType::kHifigan:
return std::make_unique<HifiganVocoder>(
config.num_threads, config.provider, config.matcha.vocoder);
case ModelType::kVocoos:
return std::make_unique<VocosVocoder>(config);
case ModelType::kUnknown:
SHERPA_ONNX_LOGE("Unknown model type in vocoder!");
return nullptr;
}
}
template <typename Manager>
std::unique_ptr<Vocoder> Vocoder::Create(Manager *mgr,
const OfflineTtsModelConfig &config) {
auto buffer = ReadFile(mgr, config.matcha.vocoder);
auto model_type = GetModelType(buffer.data(), buffer.size(), config.debug);
switch (model_type) {
case ModelType::kHifigan:
return std::make_unique<HifiganVocoder>(
config.num_threads, config.provider, config.matcha.vocoder);
case ModelType::kVocoos:
return std::make_unique<VocosVocoder>(config);
case ModelType::kUnknown:
SHERPA_ONNX_LOGE("Unknown model type in vocoder!");
return nullptr;
}
}
#if __ANDROID_API__ >= 9
template std::unique_ptr<Vocoder> Vocoder::Create(
AAssetManager *mgr, const OfflineTtsModelConfig &config);
#endif
#if __OHOS__
template std::unique_ptr<Vocoder> Vocoder::Create(
NativeResourceManager *mgr, const OfflineTtsModelConfig &config);
#endif
} // namespace sherpa_onnx
... ...
// sherpa-onnx/csrc/vocoder.h
//
// Copyright (c) 2025 Xiaomi Corporation
#ifndef SHERPA_ONNX_CSRC_VOCODER_H_
#define SHERPA_ONNX_CSRC_VOCODER_H_
#include <memory>
#include <string>
#include <vector>
#include "onnxruntime_cxx_api.h" // NOLINT
#include "sherpa-onnx/csrc/offline-tts-model-config.h"
namespace sherpa_onnx {
class Vocoder {
public:
virtual ~Vocoder() = default;
static std::unique_ptr<Vocoder> Create(const OfflineTtsModelConfig &config);
template <typename Manager>
static std::unique_ptr<Vocoder> Create(Manager *mgr,
const OfflineTtsModelConfig &config);
/** @param mel A float32 tensor of shape (batch_size, feat_dim, num_frames).
* @return Return a float32 vector containing audio samples..
*/
virtual std::vector<float> Run(Ort::Value mel) const = 0;
};
} // namespace sherpa_onnx
#endif // SHERPA_ONNX_CSRC_VOCODER_H_
... ...
// sherpa-onnx/csrc/vocos-vocoder.cc
//
// Copyright (c) 2025 Xiaomi Corporation
#include "sherpa-onnx/csrc/vocos-vocoder.h"
#include <string>
#include <utility>
#include <vector>
#if __ANDROID_API__ >= 9
#include "android/asset_manager.h"
#include "android/asset_manager_jni.h"
#endif
#if __OHOS__
#include "rawfile/raw_file_manager.h"
#endif
#include "kaldi-native-fbank/csrc/istft.h"
#include "sherpa-onnx/csrc/file-utils.h"
#include "sherpa-onnx/csrc/macros.h"
#include "sherpa-onnx/csrc/onnx-utils.h"
#include "sherpa-onnx/csrc/session.h"
namespace sherpa_onnx {
struct VocosModelMetaData {
int32_t n_fft;
int32_t hop_length;
int32_t win_length;
int32_t center;
int32_t normalized;
std::string window_type;
std::string pad_mode;
};
class VocosVocoder::Impl {
public:
explicit Impl(const OfflineTtsModelConfig &config)
: config_(config),
env_(ORT_LOGGING_LEVEL_ERROR),
sess_opts_(GetSessionOptions(config.num_threads, config.provider)),
allocator_{} {
auto buf = ReadFile(config.matcha.vocoder);
Init(buf.data(), buf.size());
}
template <typename Manager>
explicit Impl(Manager *mgr, const OfflineTtsModelConfig &config)
: config_(config),
env_(ORT_LOGGING_LEVEL_ERROR),
sess_opts_(GetSessionOptions(config.num_threads, config.provider)),
allocator_{} {
auto buf = ReadFile(mgr, config.matcha.vocoder);
Init(buf.data(), buf.size());
}
std::vector<float> Run(Ort::Value mel) const {
auto out = sess_->Run({}, input_names_ptr_.data(), &mel, 1,
output_names_ptr_.data(), output_names_ptr_.size());
std::vector<int64_t> shape = out[0].GetTensorTypeAndShapeInfo().GetShape();
if (shape[0] != 1) {
SHERPA_ONNX_LOGE("Support only batch size 1, given: %d",
static_cast<int32_t>(shape[0]));
SHERPA_ONNX_EXIT(-1);
}
knf::StftResult stft_result;
stft_result.num_frames = shape[2];
stft_result.real.resize(shape[1] * shape[2]);
stft_result.imag.resize(shape[1] * shape[2]);
// stft_result.real: (num_frames, n_fft/2+1), flattened in row major
// mag.shape: (batch_size, n_fft/2+1, num_frames)
const float *p_mag = out[0].GetTensorData<float>();
const float *p_x = out[1].GetTensorData<float>();
const float *p_y = out[2].GetTensorData<float>();
for (int32_t frame_index = 0; frame_index < static_cast<int32_t>(shape[2]);
++frame_index) {
for (int32_t bin = 0; bin < static_cast<int32_t>(shape[1]); ++bin) {
stft_result.real[frame_index * shape[1] + bin] =
p_mag[bin * shape[2] + frame_index] *
p_x[bin * shape[2] + frame_index];
stft_result.imag[frame_index * shape[1] + bin] =
p_mag[bin * shape[2] + frame_index] *
p_y[bin * shape[2] + frame_index];
}
}
knf::StftConfig stft_config;
stft_config.n_fft = meta_.n_fft;
stft_config.hop_length = meta_.hop_length;
stft_config.win_length = meta_.win_length;
stft_config.normalized = meta_.normalized;
stft_config.center = meta_.center;
stft_config.window_type = meta_.window_type;
stft_config.pad_mode = meta_.pad_mode;
knf::IStft istft(stft_config);
return istft.Compute(stft_result);
}
private:
void Init(void *model_data, size_t model_data_length) {
sess_ = std::make_unique<Ort::Session>(env_, model_data, model_data_length,
sess_opts_);
GetInputNames(sess_.get(), &input_names_, &input_names_ptr_);
GetOutputNames(sess_.get(), &output_names_, &output_names_ptr_);
// get meta data
Ort::ModelMetadata meta_data = sess_->GetModelMetadata();
if (config_.debug) {
std::ostringstream os;
os << "---Vocos model---\n";
PrintModelMetadata(os, meta_data);
os << "----------input names----------\n";
int32_t i = 0;
for (const auto &s : input_names_) {
os << i << " " << s << "\n";
++i;
}
os << "----------output names----------\n";
i = 0;
for (const auto &s : output_names_) {
os << i << " " << s << "\n";
++i;
}
#if __OHOS__
SHERPA_ONNX_LOGE("%{public}s\n", os.str().c_str());
#else
SHERPA_ONNX_LOGE("%s\n", os.str().c_str());
#endif
}
Ort::AllocatorWithDefaultOptions allocator; // used in the macro below
SHERPA_ONNX_READ_META_DATA(meta_.n_fft, "n_fft");
SHERPA_ONNX_READ_META_DATA(meta_.hop_length, "hop_length");
SHERPA_ONNX_READ_META_DATA(meta_.win_length, "win_length");
SHERPA_ONNX_READ_META_DATA(meta_.center, "center");
SHERPA_ONNX_READ_META_DATA(meta_.normalized, "normalized");
SHERPA_ONNX_READ_META_DATA_STR(meta_.window_type, "window_type");
SHERPA_ONNX_READ_META_DATA_STR(meta_.pad_mode, "pad_mode");
}
private:
OfflineTtsModelConfig config_;
VocosModelMetaData meta_;
Ort::Env env_;
Ort::SessionOptions sess_opts_;
Ort::AllocatorWithDefaultOptions allocator_;
std::unique_ptr<Ort::Session> sess_;
std::vector<std::string> input_names_;
std::vector<const char *> input_names_ptr_;
std::vector<std::string> output_names_;
std::vector<const char *> output_names_ptr_;
};
VocosVocoder::VocosVocoder(const OfflineTtsModelConfig &config)
: impl_(std::make_unique<Impl>(config)) {}
template <typename Manager>
VocosVocoder::VocosVocoder(Manager *mgr, const OfflineTtsModelConfig &config)
: impl_(std::make_unique<Impl>(mgr, config)) {}
VocosVocoder::~VocosVocoder() = default;
std::vector<float> VocosVocoder::Run(Ort::Value mel) const {
return impl_->Run(std::move(mel));
}
#if __ANDROID_API__ >= 9
template VocosVocoder::VocosVocoder(AAssetManager *mgr,
const OfflineTtsModelConfig &config);
#endif
#if __OHOS__
template VocosVocoder::VocosVocoder(NativeResourceManager *mgr,
const OfflineTtsModelConfig &config);
#endif
} // namespace sherpa_onnx
... ...
// sherpa-onnx/csrc/vocos-vocoder.h
//
// Copyright (c) 2025 Xiaomi Corporation
#ifndef SHERPA_ONNX_CSRC_VOCOS_VOCODER_H_
#define SHERPA_ONNX_CSRC_VOCOS_VOCODER_H_
#include <memory>
#include <string>
#include <vector>
#include "onnxruntime_cxx_api.h" // NOLINT
#include "sherpa-onnx/csrc/offline-tts-model-config.h"
#include "sherpa-onnx/csrc/vocoder.h"
namespace sherpa_onnx {
class VocosVocoder : public Vocoder {
public:
~VocosVocoder() override;
explicit VocosVocoder(const OfflineTtsModelConfig &config);
template <typename Manager>
VocosVocoder(Manager *mgr, const OfflineTtsModelConfig &config);
/** @param mel A float32 tensor of shape (batch_size, feat_dim, num_frames).
* @return Return a float32 tensor of shape (batch_size, num_samples).
*/
std::vector<float> Run(Ort::Value mel) const override;
private:
class Impl;
std::unique_ptr<Impl> impl_;
};
} // namespace sherpa_onnx
#endif // SHERPA_ONNX_CSRC_VOCOS_VOCODER_H_
... ...
... ... @@ -17,8 +17,8 @@ if [ ! -f ./matcha-icefall-en_US-ljspeech/model-steps-3.onnx ]; then
rm matcha-icefall-en_US-ljspeech.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
if [ ! -e ./tts-matcha-en ]; then
... ...
... ... @@ -16,8 +16,8 @@ if [ ! -f ./matcha-icefall-zh-baker/model-steps-3.onnx ]; then
rm matcha-icefall-zh-baker.tar.bz2
fi
if [ ! -f ./hifigan_v2.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/hifigan_v2.onnx
if [ ! -f ./vocos-22khz-univ.onnx ]; then
curl -SL -O https://github.com/k2-fsa/sherpa-onnx/releases/download/vocoder-models/vocos-22khz-univ.onnx
fi
if [ ! -e ./tts-matcha-zh ]; then
... ...
... ... @@ -6,7 +6,7 @@ class MyClass {
func run() {
let acousticModel = "./matcha-icefall-en_US-ljspeech/model-steps-3.onnx"
let vocoder = "./hifigan_v2.onnx"
let vocoder = "./vocos-22khz-univ.onnx"
let tokens = "./matcha-icefall-en_US-ljspeech/tokens.txt"
let dataDir = "./matcha-icefall-en_US-ljspeech/espeak-ng-data"
let matcha = sherpaOnnxOfflineTtsMatchaModelConfig(
... ...
... ... @@ -6,7 +6,7 @@ class MyClass {
func run() {
let acousticModel = "./matcha-icefall-zh-baker/model-steps-3.onnx"
let vocoder = "./hifigan_v2.onnx"
let vocoder = "./vocos-22khz-univ.onnx"
let lexicon = "./matcha-icefall-zh-baker/lexicon.txt"
let tokens = "./matcha-icefall-zh-baker/tokens.txt"
let dictDir = "./matcha-icefall-zh-baker/dict"
... ...