From be4aedd8e5d365b0542e132ed8f0f8f5fbe93ddf Mon Sep 17 00:00:00 2001
From: Fangjun Kuang <csukuangfj@gmail.com>
Date: Sat, 14 Sep 2024 12:08:56 +0800
Subject: [PATCH] Update Pascal API to support max speech duration in VAD

---
 .github/workflows/dot-net.yaml                           | 2 ++
 java-api-examples/VadRemoveSilence.java                  | 1 +
 lazarus-examples/generate_subtitles/my_init.pas          | 3 ++-
 .../vad-remove-non-speech-segments-from-file.py          | 9 +++++++++
 sherpa-onnx/pascal-api/sherpa_onnx.pas                   | 9 +++++++--
 5 files changed, 21 insertions(+), 3 deletions(-)

diff --git a/.github/workflows/dot-net.yaml b/.github/workflows/dot-net.yaml
index 5299b19ce..36637a9e2 100644
--- a/.github/workflows/dot-net.yaml
+++ b/.github/workflows/dot-net.yaml
@@ -93,6 +93,8 @@ jobs:
             git clone https://huggingface.co/csukuangfj/sherpa-onnx-libs huggingface
 
             cd huggingface
+            git fetch
+            git pull
             mkdir -p windows-for-dotnet
 
             cp -v ../sherpa-onnx-*.tar.bz2 ./windows-for-dotnet
diff --git a/java-api-examples/VadRemoveSilence.java b/java-api-examples/VadRemoveSilence.java
index 3af1caa7f..511a508e4 100644
--- a/java-api-examples/VadRemoveSilence.java
+++ b/java-api-examples/VadRemoveSilence.java
@@ -19,6 +19,7 @@ public static void main(String[] args) {
             .setMinSilenceDuration(0.25f)
             .setMinSpeechDuration(0.5f)
             .setWindowSize(512)
+            .setMaxSpeechDuration(5.0f)
             .build();
 
     VadModelConfig config =
diff --git a/lazarus-examples/generate_subtitles/my_init.pas b/lazarus-examples/generate_subtitles/my_init.pas
index 55df79f15..d57448b6d 100644
--- a/lazarus-examples/generate_subtitles/my_init.pas
+++ b/lazarus-examples/generate_subtitles/my_init.pas
@@ -48,8 +48,9 @@ function CreateVad(VadFilename: AnsiString): TSherpaOnnxVoiceActivityDetector;
   WindowSize := 512; {Please don't change it unless you know the details}
 
   Config.SileroVad.Model := VadFilename;
-  Config.SileroVad.MinSpeechDuration := 0.5;
+  Config.SileroVad.MinSpeechDuration := 0.25;
   Config.SileroVad.MinSilenceDuration := 0.5;
+  Config.SileroVad.MaxSpeechDuration := 5.0;
   Config.SileroVad.Threshold := 0.5;
   Config.SileroVad.WindowSize := WindowSize;
   Config.NumThreads:= 2;
diff --git a/python-api-examples/vad-remove-non-speech-segments-from-file.py b/python-api-examples/vad-remove-non-speech-segments-from-file.py
index f559e6519..ad4814487 100755
--- a/python-api-examples/vad-remove-non-speech-segments-from-file.py
+++ b/python-api-examples/vad-remove-non-speech-segments-from-file.py
@@ -90,6 +90,15 @@ def main():
 
     config = sherpa_onnx.VadModelConfig()
     config.silero_vad.model = args.silero_vad_model
+    config.silero_vad.threshold = 0.5
+    config.silero_vad.min_silence_duration = 0.25  # seconds
+    config.silero_vad.min_speech_duration = 0.25  # seconds
+
+    # If the current segment is larger than this value, then it increases
+    # the threshold to 0.9 internally. After detecting this segment,
+    # it resets the threshold to its original value.
+    config.silero_vad.max_speech_duration = 5  # seconds
+
     config.sample_rate = sample_rate
 
     window_size = config.silero_vad.window_size
diff --git a/sherpa-onnx/pascal-api/sherpa_onnx.pas b/sherpa-onnx/pascal-api/sherpa_onnx.pas
index 987b31f14..7f05793e1 100644
--- a/sherpa-onnx/pascal-api/sherpa_onnx.pas
+++ b/sherpa-onnx/pascal-api/sherpa_onnx.pas
@@ -341,6 +341,7 @@   TSherpaOnnxSileroVadModelConfig = record
     MinSilenceDuration: Single;
     MinSpeechDuration: Single;
     WindowSize: Integer;
+    MaxSpeechDuration: Single;
     function ToString: AnsiString;
     class operator Initialize({$IFDEF FPC}var{$ELSE}out{$ENDIF} Dest: TSherpaOnnxSileroVadModelConfig);
   end;
@@ -594,6 +595,7 @@   SherpaOnnxSileroVadModelConfig = record
     MinSilenceDuration: cfloat;
     MinSpeechDuration: cfloat;
     WindowSize: cint32;
+    MaxSpeechDuration: cfloat;
   end;
   SherpaOnnxVadModelConfig = record
     SileroVad: SherpaOnnxSileroVadModelConfig;
@@ -1402,10 +1404,11 @@ function TSherpaOnnxSileroVadModelConfig.ToString: AnsiString;
     'Threshold := %.2f, ' +
     'MinSilenceDuration := %.2f, ' +
     'MinSpeechDuration := %.2f, ' +
-    'WindowSize := %d' +
+    'WindowSize := %d, ' +
+    'MaxSpeechDuration := %.2f' +
     ')',
     [Self.Model, Self.Threshold, Self.MinSilenceDuration,
-     Self.MinSpeechDuration, Self.WindowSize
+     Self.MinSpeechDuration, Self.WindowSize, Self.MaxSpeechDuration
     ]);
 end;
 
@@ -1415,6 +1418,7 @@ function TSherpaOnnxSileroVadModelConfig.ToString: AnsiString;
   Dest.MinSilenceDuration := 0.5;
   Dest.MinSpeechDuration := 0.25;
   Dest.WindowSize := 512;
+  Dest.MaxSpeechDuration := 5.0;
 end;
 
 function TSherpaOnnxVadModelConfig.ToString: AnsiString;
@@ -1569,6 +1573,7 @@ constructor TSherpaOnnxVoiceActivityDetector.Create(Config: TSherpaOnnxVadModelC
   C.SileroVad.MinSilenceDuration := Config.SileroVad.MinSilenceDuration;
   C.SileroVad.MinSpeechDuration := Config.SileroVad.MinSpeechDuration;
   C.SileroVad.WindowSize := Config.SileroVad.WindowSize;
+  C.SileroVad.MaxSpeechDuration := Config.SileroVad.MaxSpeechDuration;
 
   C.SampleRate := Config.SampleRate;
   C.NumThreads := Config.NumThreads;