Add C API for offline TTS. (#373)

k2-fsa · Oct 19, 2023 · ea7c45b · ea7c45b
1 parent 58ab7e7
commit ea7c45b
Show file tree

Hide file tree

Showing 16 changed files with 562 additions and 24 deletions.
diff --git a/.github/scripts/test-offline-tts.sh b/.github/scripts/test-offline-tts.sh
@@ -0,0 +1,98 @@
+#!/usr/bin/env bash
+
+set -e
+
+log() {
+  # This function is from espnet
+  local fname=${BASH_SOURCE[1]##*/}
+  echo -e "$(date '+%Y-%m-%d %H:%M:%S') (${fname}:${BASH_LINENO[0]}:${FUNCNAME[1]}) $*"
+}
+
+echo "EXE is $EXE"
+echo "PATH: $PATH"
+
+which $EXE
+
+# test waves are saved in ./tts
+mkdir ./tts
+
+log "------------------------------------------------------------"
+log "vits-ljs test"
+log "------------------------------------------------------------"
+
+repo_url=https://huggingface.co/csukuangfj/vits-ljs
+log "Start testing ${repo_url}"
+repo=$(basename $repo_url)
+log "Download pretrained model from $repo_url"
+GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+pushd $repo
+git lfs pull --include "*.onnx"
+ls -lh *.onnx
+popd
+
+$EXE \
+  --vits-model=$repo/vits-ljs.onnx \
+  --vits-lexicon=$repo/lexicon.txt \
+  --vits-tokens=$repo/tokens.txt \
+  --output-filename=./tts/vits-ljs.wav \
+  'liliana, the most beautiful and lovely assistant of our team!'
+
+ls -lh ./tts
+
+rm -rfv $repo
+
+log "------------------------------------------------------------"
+log "vits-vctk test"
+log "------------------------------------------------------------"
+
+repo_url=https://huggingface.co/csukuangfj/vits-vctk
+log "Start testing ${repo_url}"
+repo=$(basename $repo_url)
+log "Download pretrained model from $repo_url"
+GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+pushd $repo
+git lfs pull --include "*.onnx"
+ls -lh *.onnx
+popd
+
+for sid in 0 10 90; do
+  $EXE \
+    --vits-model=$repo/vits-vctk.onnx \
+    --vits-lexicon=$repo/lexicon.txt \
+    --vits-tokens=$repo/tokens.txt \
+    --sid=$sid \
+    --output-filename=./tts/vits-vctk-${sid}.wav \
+    'liliana, the most beautiful and lovely assistant of our team!'
+done
+
+rm -rfv $repo
+
+ls -lh tts/
+
+log "------------------------------------------------------------"
+log "vits-zh-aishell3"
+log "------------------------------------------------------------"
+
+repo_url=https://huggingface.co/csukuangfj/vits-zh-aishell3
+log "Start testing ${repo_url}"
+repo=$(basename $repo_url)
+log "Download pretrained model from $repo_url"
+GIT_LFS_SKIP_SMUDGE=1 git clone $repo_url
+pushd $repo
+git lfs pull --include "*.onnx"
+ls -lh *.onnx
+popd
+
+for sid in 0 10 90; do
+  $EXE \
+    --vits-model=$repo/vits-aishell3.onnx \
+    --vits-lexicon=$repo/lexicon.txt \
+    --vits-tokens=$repo/tokens.txt \
+    --sid=$sid \
+    --output-filename=./tts/vits-aishell3-${sid}.wav \
+    '林美丽最美丽'
+done
+
+rm -rfv $repo
+
+ls -lh ./tts/
diff --git a/.github/workflows/linux-gpu.yaml b/.github/workflows/linux-gpu.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -26,6 +27,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -48,7 +50,7 @@ jobs:
         build_type: [Release, Debug]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           fetch-depth: 0
 
@@ -75,6 +77,14 @@ jobs:
           file build/bin/sherpa-onnx
           readelf -d build/bin/sherpa-onnx
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin:$PATH
+          export EXE=sherpa-onnx-offline-tts
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer
         shell: bash
         run: |

diff --git a/.github/workflows/linux.yaml b/.github/workflows/linux.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -26,6 +27,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -49,7 +51,7 @@ jobs:
         shared_lib: [ON, OFF]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           fetch-depth: 0
 
@@ -76,6 +78,14 @@ jobs:
           file build/bin/sherpa-onnx
           readelf -d build/bin/sherpa-onnx
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin:$PATH
+          export EXE=sherpa-onnx-offline-tts
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer
         shell: bash
         run: |
@@ -150,3 +160,8 @@ jobs:
           file_glob: true
           overwrite: true
           file: sherpa-onnx-*linux-x64.tar.bz2
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: tts-generated-test-files
+          path: tts
diff --git a/.github/workflows/macos.yaml b/.github/workflows/macos.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -24,6 +25,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -44,7 +46,7 @@ jobs:
         build_type: [Release, Debug]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           fetch-depth: 0
 
@@ -74,6 +76,14 @@ jobs:
           otool -L build/bin/sherpa-onnx
           otool -l build/bin/sherpa-onnx
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin:$PATH
+          export EXE=sherpa-onnx-offline-tts
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer
         shell: bash
         run: |

diff --git a/.github/workflows/pkg-config.yaml b/.github/workflows/pkg-config.yaml
@@ -9,6 +9,7 @@ on:
       - '*'
     paths:
       - '.github/workflows/pkg-config.yaml'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -19,6 +20,7 @@ on:
       - master
     paths:
       - '.github/workflows/pkg-config.yaml'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -92,13 +94,29 @@ jobs:
         run: |
           export PKG_CONFIG_PATH=$PWD/build/install:$PKG_CONFIG_PATH
           cd c-api-examples
+
           gcc -o decode-file-c-api $(pkg-config --cflags sherpa-onnx) ./decode-file-c-api.c $(pkg-config --libs sherpa-onnx)
           ./decode-file-c-api --help
 
+          gcc -o offline-tts-c-api $(pkg-config --cflags sherpa-onnx) ./offline-tts-c-api.c $(pkg-config --libs sherpa-onnx)
+          ./offline-tts-c-api --help
+
+      - name: Test offline TTS C API
+        shell: bash
+        run: |
+          export PATH=$PWD/c-api-examples:$PATH
+          export EXE=offline-tts-c-api
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online transducer (C API)
         shell: bash
         run: |
           export PATH=$PWD/c-api-examples:$PATH
           export EXE=decode-file-c-api
 
           .github/scripts/test-online-transducer.sh
+
+      - uses: actions/upload-artifact@v3
+        with:
+          name: tts-generated-test-files
+          path: tts
diff --git a/.github/workflows/windows-x64-cuda.yaml b/.github/workflows/windows-x64-cuda.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -24,6 +25,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -64,6 +66,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx-offline-tts.exe
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer for windows x64
         shell: bash
         run: |

diff --git a/.github/workflows/windows-x64.yaml b/.github/workflows/windows-x64.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -24,6 +25,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -45,7 +47,7 @@ jobs:
         shared_lib: [ON, OFF]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           fetch-depth: 0
 
@@ -65,6 +67,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx-offline-tts.exe
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer for windows x64
         shell: bash
         run: |

diff --git a/.github/workflows/windows-x86.yaml b/.github/workflows/windows-x86.yaml
@@ -12,6 +12,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -24,6 +25,7 @@ on:
       - '.github/scripts/test-online-paraformer.sh'
       - '.github/scripts/test-offline-transducer.sh'
       - '.github/scripts/test-offline-ctc.sh'
+      - '.github/scripts/test-offline-tts.sh'
       - 'CMakeLists.txt'
       - 'cmake/**'
       - 'sherpa-onnx/csrc/*'
@@ -45,7 +47,7 @@ jobs:
         shared_lib: [ON, OFF]
 
     steps:
-      - uses: actions/checkout@v2
+      - uses: actions/checkout@v3
         with:
           fetch-depth: 0
 
@@ -65,6 +67,14 @@ jobs:
 
           ls -lh ./bin/Release/sherpa-onnx.exe
 
+      - name: Test offline TTS
+        shell: bash
+        run: |
+          export PATH=$PWD/build/bin/Release:$PATH
+          export EXE=sherpa-onnx-offline-tts.exe
+
+          .github/scripts/test-offline-tts.sh
+
       - name: Test online paraformer for windows x86
         shell: bash
         run: |

diff --git a/.gitignore b/.gitignore
@@ -21,6 +21,7 @@ run-bilingual*.sh
 run-*-zipformer.sh
 run-zh.sh
 decode-file-c-api
+offline-tts-c-api
 run-decode-file-c-api.sh
 sherpa-onnx-ffmpeg
 build-ios

diff --git a/c-api-examples/CMakeLists.txt b/c-api-examples/CMakeLists.txt
@@ -3,3 +3,6 @@ include(cargs)
 include_directories(${CMAKE_SOURCE_DIR})
 add_executable(decode-file-c-api decode-file-c-api.c)
 target_link_libraries(decode-file-c-api sherpa-onnx-c-api cargs)
+
+add_executable(offline-tts-c-api offline-tts-c-api.c)
+target_link_libraries(offline-tts-c-api sherpa-onnx-c-api cargs)