Add Lazarus example for Moonshine models. (#1532)

k2-fsa · Nov 12, 2024 · a16c9af · a16c9af
1 parent 4fab3f2
commit a16c9af
Show file tree

Hide file tree

Showing 4 changed files with 78 additions and 8 deletions.
diff --git a/build-ios-shared.sh b/build-ios-shared.sh
@@ -21,6 +21,8 @@ fi
 
 if [ ! -z CMAKE_VERBOSE_MAKEFILE ]; then
   CMAKE_VERBOSE_MAKEFILE=ON
+else
+  CMAKE_VERBOSE_MAKEFILE=OFF
 fi
 
 if [ ! -f $onnxruntime_dir/onnxruntime.xcframework/ios-arm64/onnxruntime.a ]; then

diff --git a/cmake/piper-phonemize.cmake b/cmake/piper-phonemize.cmake
@@ -1,18 +1,18 @@
 function(download_piper_phonemize)
   include(FetchContent)
 
-  set(piper_phonemize_URL  "https://github.com/csukuangfj/piper-phonemize/archive/38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip")
-  set(piper_phonemize_URL2 "https://hf-mirror.com/csukuangfj/sherpa-onnx-cmake-deps/resolve/main/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip")
-  set(piper_phonemize_HASH "SHA256=ab4d06ca76047e1585c63c482f39ffead5315785345055360703cc9382c5e74b")
+  set(piper_phonemize_URL  "https://github.com/csukuangfj/piper-phonemize/archive/78a788e0b719013401572d70fef372e77bff8e43.zip")
+  set(piper_phonemize_URL2 "https://hf-mirror.com/csukuangfj/sherpa-onnx-cmake-deps/resolve/main/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip")
+  set(piper_phonemize_HASH "SHA256=89641a46489a4898754643ce57bda9c9b54b4ca46485fdc02bf0dc84b866645d")
 
   # If you don't have access to the Internet,
   # please pre-download kaldi-decoder
   set(possible_file_locations
-    $ENV{HOME}/Downloads/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip
-    ${CMAKE_SOURCE_DIR}/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip
-    ${CMAKE_BINARY_DIR}/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip
-    /tmp/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip
-    /star-fj/fangjun/download/github/piper-phonemize-38ee199dcc49c7b6de89f7ebfb32ed682763fa1b.zip
+    $ENV{HOME}/Downloads/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip
+    ${CMAKE_SOURCE_DIR}/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip
+    ${CMAKE_BINARY_DIR}/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip
+    /tmp/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip
+    /star-fj/fangjun/download/github/piper-phonemize-78a788e0b719013401572d70fef372e77bff8e43.zip
   )
 
   foreach(f IN LISTS possible_file_locations)

diff --git a/lazarus-examples/generate_subtitles/my_init.pas b/lazarus-examples/generate_subtitles/my_init.pas
@@ -159,6 +159,30 @@ function CreateOfflineRecognizerWhisper(
   Result := TSherpaOnnxOfflineRecognizer.Create(Config);
 end;
 
+function CreateOfflineRecognizerMoonshine(
+  Tokens: AnsiString;
+  Preprocessor: AnsiString;
+  Encoder: AnsiString;
+  UncachedDecoder: AnsiString;
+  CachedDecoder: AnsiString): TSherpaOnnxOfflineRecognizer;
+var
+  Config: TSherpaOnnxOfflineRecognizerConfig;
+begin
+  Initialize(Config);
+
+  Config.ModelConfig.Moonshine.Preprocessor := Preprocessor;
+  Config.ModelConfig.Moonshine.Encoder := Encoder;
+  Config.ModelConfig.Moonshine.UncachedDecoder := UncachedDecoder;
+  Config.ModelConfig.Moonshine.CachedDecoder := CachedDecoder;
+
+  Config.ModelConfig.Tokens := Tokens;
+  Config.ModelConfig.Provider := 'cpu';
+  Config.ModelConfig.NumThreads := 2;
+  Config.ModelConfig.Debug := False;
+
+  Result := TSherpaOnnxOfflineRecognizer.Create(Config);
+end;
+
 constructor TMyInitThread.Create(CreateSuspended : boolean; ModelDirectory: AnsiString);
 begin
   inherited Create(CreateSuspended);
@@ -193,6 +217,11 @@ procedure TMyInitThread.Execute;
   NeMoTransducerEncoder: AnsiString;
   NeMoTransducerDecoder: AnsiString;
   NeMoTransducerJoiner: AnsiString;
+
+  MoonshinePreprocessor: AnsiString;
+  MoonshineEncoder: AnsiString;
+  MoonshineUncachedDecoder: AnsiString;
+  MoonshineCachedDecoder: AnsiString;
 begin
     VadFilename := ModelDir + 'silero_vad.onnx';
     Tokens := ModelDir + 'tokens.txt';
@@ -292,6 +321,24 @@ procedure TMyInitThread.Execute;
     NeMoTransducerDecoder := ModelDir + 'nemo-transducer-decoder.onnx';
     NeMoTransducerJoiner := ModelDir + 'nemo-transducer-joiner.onnx';
 
+    {
+      Please Visit
+      https://k2-fsa.github.io/sherpa/onnx/moonshine/models.html
+      to download a Moonshine model.
+
+      Note that you have to rename model files after downloading. The following
+      is an example.
+
+      mv preprocess.onnx moonshine-preprocessor.onnx
+      mv encode.int8.onnx moonshine-encoder.onnx
+      mv uncached_decode.int8.onnx moonshine-uncached-decoder.onnx
+      mv cached_decode.int8.onnx moonshine-cached-decoder.onnx
+    }
+    MoonshinePreprocessor := ModelDir + 'moonshine-preprocessor.onnx';
+    MoonshineEncoder := ModelDir + 'moonshine-encoder.onnx';
+    MoonshineUncachedDecoder := ModelDir + 'moonshine-uncached-decoder.onnx';
+    MoonshineCachedDecoder := ModelDir + 'moonshine-cached-decoder.onnx';
+
     if not FileExists(VadFilename) then
       begin
         Status := VadFilename + ' does not exist! Please download it from' +
@@ -344,6 +391,13 @@ procedure TMyInitThread.Execute;
             NeMoTransducerEncoder, NeMoTransducerDecoder, NeMoTransducerJoiner, 'nemo_transducer');
           Msg := 'NeMo transducer';
         end
+    else if FileExists(MoonshinePreprocessor) and FileExists(MoonshineEncoder) and FileExists(MoonshineUncachedDecoder) and FileExists(MoonshineCachedDecoder) then
+        begin
+          Form1.OfflineRecognizer := CreateOfflineRecognizerMoonshine(Tokens,
+            MoonshinePreprocessor, MoonshineEncoder, MoonshineUncachedDecoder,
+            MoonshineCachedDecoder);
+          Msg := 'Moonshine';
+        end
     else
       begin
         Status := 'Please download at least one non-streaming speech recognition model first.';

diff --git a/scripts/lazarus/generate-subtitles.py b/scripts/lazarus/generate-subtitles.py
@@ -50,6 +50,20 @@ def get_models():
             popd
             """,
         ),
+        Model(
+            model_name="sherpa-onnx-moonshine-tiny-en-int8",
+            lang="en",
+            short_name="moonshine_tiny",
+            cmd="""
+            pushd $model_name
+            mv -v preprocess.onnx moonshine-preprocessor.onnx
+            mv -v encode.int8.onnx moonshine-encoder.onnx
+            mv -v uncached_decode.int8.onnx moonshine-uncached-decoder.onnx
+            mv -v cached_decode.int8.onnx moonshine-cached-decoder.onnx
+
+            popd
+            """,
+        ),
         Model(
             model_name="sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17",
             lang="zh_en_ko_ja_yue",