Add more Chinese TTS models (Mandarin and Cantonese) (#589)

2024-02-20 15:05:35 +08:00
parent 5f075d0fce
commit d2cc48ded5
25 changed files with 131 additions and 69 deletions
--- a/.github/workflows/apk-kws.yaml
+++ b/.github/workflows/apk-kws.yaml
@@ -29,6 +29,12 @@ jobs:
        with:
          fetch-depth: 0

+      # https://github.com/actions/setup-java
+      - uses: actions/setup-java@v4
+        with:
+          distribution: 'temurin' # See 'Supported distributions' for available options
+          java-version: '21'
+
      - name: ccache
        uses: hendrikmuhs/ccache-action@v1.2
        with:
--- a/.github/workflows/apk-speaker-identification.yaml
+++ b/.github/workflows/apk-speaker-identification.yaml
@@ -98,7 +98,7 @@ jobs:
        if: true
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/apk-tts-engine.yaml
+++ b/.github/workflows/apk-tts-engine.yaml
@@ -94,7 +94,7 @@ jobs:
        if: true
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/apk-tts.yaml
+++ b/.github/workflows/apk-tts.yaml
@@ -34,6 +34,12 @@ jobs:
        with:
          fetch-depth: 0

+      # https://github.com/actions/setup-java
+      - uses: actions/setup-java@v4
+        with:
+          distribution: 'temurin' # See 'Supported distributions' for available options
+          java-version: '21'
+
      - name: ccache
        uses: hendrikmuhs/ccache-action@v1.2
        with:
@@ -89,7 +95,7 @@ jobs:
        if: true
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/apk.yaml
+++ b/.github/workflows/apk.yaml
@@ -29,6 +29,12 @@ jobs:
        with:
          fetch-depth: 0

+      # https://github.com/actions/setup-java
+      - uses: actions/setup-java@v4
+        with:
+          distribution: 'temurin' # See 'Supported distributions' for available options
+          java-version: '21'
+
      - name: ccache
        uses: hendrikmuhs/ccache-action@v1.2
        with:
--- a/.github/workflows/build-wheels-aarch64.yaml
+++ b/.github/workflows/build-wheels-aarch64.yaml
@@ -63,7 +63,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-armv7l.yaml
+++ b/.github/workflows/build-wheels-armv7l.yaml
@@ -87,7 +87,7 @@ jobs:
        if: matrix.python-version == '3.8'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-linux.yaml
+++ b/.github/workflows/build-wheels-linux.yaml
@@ -79,7 +79,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-macos-arm64.yaml
+++ b/.github/workflows/build-wheels-macos-arm64.yaml
@@ -68,7 +68,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-macos-x64.yaml
+++ b/.github/workflows/build-wheels-macos-x64.yaml
@@ -68,7 +68,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-win32.yaml
+++ b/.github/workflows/build-wheels-win32.yaml
@@ -53,7 +53,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/build-wheels-win64.yaml
+++ b/.github/workflows/build-wheels-win64.yaml
@@ -52,7 +52,7 @@ jobs:
        if: matrix.python-version == 'cp38'
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/.github/workflows/export-wenet-to-onnx.yaml
+++ b/.github/workflows/export-wenet-to-onnx.yaml
@@ -36,7 +36,7 @@ jobs:
      - name: Publish to huggingface (aishell)
        env:
          HF_TOKEN: ${{ secrets.HF_TOKEN }}
-        uses: nick-fields/retry@v2
+        uses: nick-fields/retry@v3
        with:
          max_attempts: 20
          timeout_seconds: 200
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -117,14 +117,14 @@ message(STATUS "C++ Standard version: ${CMAKE_CXX_STANDARD}")

 include(CheckIncludeFileCXX)

-if(UNIX AND NOT APPLE AND NOT SHERPA_ONNX_ENABLE_WASM)
+if(UNIX AND NOT APPLE AND NOT SHERPA_ONNX_ENABLE_WASM AND NOT CMAKE_SYSTEM_NAME STREQUAL Android)
  check_include_file_cxx(alsa/asoundlib.h SHERPA_ONNX_HAS_ALSA)
  if(SHERPA_ONNX_HAS_ALSA)
    add_definitions(-DSHERPA_ONNX_ENABLE_ALSA=1)
  else()
    message(WARNING "\
 Could not find alsa/asoundlib.h !
-We won't build sherpa-ncnn-alsa
+We won't build sherpa-onnx-alsa
 To fix that, please do:
  (1) sudo apt-get install alsa-utils libasound2-dev
  (2) rm -rf build
--- a/android/SherpaOnnx/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnx/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Thu Feb 23 11:09:06 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnx2Pass/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnx2Pass/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Sun Sep 10 18:03:03 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnxKws/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnxKws/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Thu Feb 23 11:09:06 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnxTts/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnxTts/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Mon Oct 23 15:40:58 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnxVad/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnxVad/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Sat Sep 23 10:24:21 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnxVadAsr/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnxVadAsr/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Sat Sep 23 20:50:52 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/android/SherpaOnnxWebSocket/gradle/wrapper/gradle-wrapper.properties
+++ b/android/SherpaOnnxWebSocket/gradle/wrapper/gradle-wrapper.properties
@@ -1,6 +1,6 @@
 #Thu Feb 23 11:09:06 CST 2023
 distributionBase=GRADLE_USER_HOME
-distributionUrl=https\://services.gradle.org/distributions/gradle-7.4-bin.zip
+distributionUrl=https\://services.gradle.org/distributions/gradle-8.2-bin.zip
 distributionPath=wrapper/dists
 zipStorePath=wrapper/dists
 zipStoreBase=GRADLE_USER_HOME
--- a/scripts/apk/build-apk-speaker-identification.sh.in
+++ b/scripts/apk/build-apk-speaker-identification.sh.in
@@ -65,6 +65,8 @@ for arch in arm64-v8a armeabi-v7a x86_64 x86; do
  cp -v ./build-android-$src_arch/install/lib/*.so ./android/SherpaOnnxSpeakerIdentification/app/src/main/jniLibs/$arch/

  pushd ./android/SherpaOnnxSpeakerIdentification
+  sed -i.bak s/2048/9012/g ./gradle.properties
+  git diff ./gradle.properties
  ./gradlew build
  popd

--- a/scripts/apk/build-apk-tts-engine.sh.in
+++ b/scripts/apk/build-apk-tts-engine.sh.in
@@ -82,6 +82,8 @@ for arch in arm64-v8a armeabi-v7a x86_64 x86; do
  cp -v ./build-android-$src_arch/install/lib/*.so ./android/SherpaOnnxTtsEngine/app/src/main/jniLibs/$arch/

  pushd ./android/SherpaOnnxTtsEngine
+  sed -i.bak s/2048/9012/g ./gradle.properties
+  git diff ./gradle.properties
  ./gradlew build
  popd

--- a/scripts/apk/build-apk-tts.sh.in
+++ b/scripts/apk/build-apk-tts.sh.in
@@ -80,6 +80,8 @@ for arch in arm64-v8a armeabi-v7a x86_64 x86; do
  cp -v ./build-android-$src_arch/install/lib/*.so ./android/SherpaOnnxTts/app/src/main/jniLibs/$arch/

  pushd ./android/SherpaOnnxTts
+  sed -i.bak s/2048/9012/g ./gradle.properties
+  git diff ./gradle.properties
  ./gradlew build
  popd

--- a/scripts/apk/generate-tts-apk-script.py
+++ b/scripts/apk/generate-tts-apk-script.py
@@ -39,7 +39,8 @@ class TtsModel:

 def convert_lang_to_iso_639_3(models: List[TtsModel]):
    for m in models:
-        m.lang_iso_639_3 = Lang(m.lang).pt3
+        if m.lang_iso_639_3 == "":
+            m.lang_iso_639_3 = Lang(m.lang).pt3


 def get_coqui_models() -> List[TtsModel]:
@@ -209,54 +210,91 @@ def get_vits_models() -> List[TtsModel]:
            lang="zh",
            rule_fsts="vits-zh-aishell3/rule.fst",
        ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-doom",
-        #      model_name="doom.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-doom/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-echo",
-        #      model_name="echo.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-echo/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-zenyatta",
-        #      model_name="zenyatta.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-zenyatta/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-abyssinvoker",
-        #      model_name="abyssinvoker.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-abyssinvoker/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-keqing",
-        #      model_name="keqing.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-keqing/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-eula",
-        #      model_name="eula.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-eula/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-bronya",
-        #      model_name="bronya.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-bronya/rule.fst",
-        #  ),
-        #  TtsModel(
-        #      model_dir="vits-zh-hf-theresa",
-        #      model_name="theresa.onnx",
-        #      lang="zh",
-        #      rule_fsts="vits-zh-hf-theresa/rule.fst",
-        #  ),
+        TtsModel(
+            model_dir="vits-zh-hf-doom",
+            model_name="doom.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-doom/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-echo",
+            model_name="echo.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-echo/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-zenyatta",
+            model_name="zenyatta.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-zenyatta/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-abyssinvoker",
+            model_name="abyssinvoker.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-abyssinvoker/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-keqing",
+            model_name="keqing.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-keqing/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-eula",
+            model_name="eula.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-eula/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-bronya",
+            model_name="bronya.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-bronya/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-theresa",
+            model_name="theresa.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-theresa/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-fanchen-wnj",
+            model_name="vits-zh-hf-fanchen-wnj.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-fanchen-wnj/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-fanchen-C",
+            model_name="vits-zh-hf-fanchen-C.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-fanchen-C/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-fanchen-ZhiHuiLaoZhe",
+            model_name="vits-zh-hf-fanchen-ZhiHuiLaoZhe.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-fanchen-ZhiHuiLaoZhe/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-fanchen-ZhiHuiLaoZhe_new",
+            model_name="vits-zh-hf-fanchen-ZhiHuiLaoZhe_new.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-fanchen-ZhiHuiLaoZhe_new/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-zh-hf-fanchen-unity",
+            model_name="vits-zh-hf-fanchen-unity.onnx",
+            lang="zh",
+            rule_fsts="vits-zh-hf-fanchen-unity/rule.fst",
+        ),
+        TtsModel(
+            model_dir="vits-cantonese-hf-xiaomaiiwn",
+            model_name="vits-cantonese-hf-xiaomaiiwn.onnx",
+            lang="cantonese",
+            lang_iso_639_3="yue",
+            rule_fsts="vits-cantonese-hf-xiaomaiiwn/rule.fst",
+        ),
        # English (US)
        TtsModel(model_dir="vits-vctk", model_name="vits-vctk.onnx", lang="en"),
        #  TtsModel(model_dir="vits-ljs", model_name="vits-ljs.onnx", lang="en"),