add endpoint for websocket server

sherpa-onnx/csrc/online-recognizer.cc

@@ -244,18 +244,40 @@ class OnlineRecognizer::Impl {
     // TODO(fangjun): Remember to change these constants if needed
     int32_t frame_shift_ms = 10;
     int32_t subsampling_factor = 4;
-    return Convert(decoder_result, sym_, frame_shift_ms, subsampling_factor);
+    auto ans = Convert(decoder_result,
+               sym_,
+               frame_shift_ms,
+               subsampling_factor);
+
+    if (!IsReady(s) && s->IsLastFrame(s->NumFramesReady() - 1)) {
+      ans.is_final = true;
+    }
+    ans.segment = s->GetWavSegment();
+    float frame_shift_s = frame_shift_ms / 1000.;
+    ans.start_time = s->GetStartFrame() * frame_shift_s;
+    s->GetNumTrailingBlankFrames() = decoder_result.num_trailing_blanks;
+
+    if (config_.enable_endpoint && IsEndpoint(s)) {
+      auto r = decoder_->GetEmptyResult();
+      s->SetResult(r);
+      ans.is_final = true;
+      s->GetWavSegment() += 1;
+      s->GetStartFrame() = s->GetNumProcessedFrames();
+      s->GetNumTrailingBlankFrames() = 0;
+    }
+    return ans;
   }
 
   bool IsEndpoint(OnlineStream *s) const {
     if (!config_.enable_endpoint) return false;
-    int32_t num_processed_frames = s->GetNumProcessedFrames();
+    int32_t num_processed_frames = s->GetNumProcessedFrames() -
+                                   s->GetStartFrame();
 
     // frame shift is 10 milliseconds
     float frame_shift_in_seconds = 0.01;
 
     // subsampling factor is 4
-    int32_t trailing_silence_frames = s->GetResult().num_trailing_blanks * 4;
+    int32_t trailing_silence_frames = s->GetNumTrailingBlankFrames() * 4;
 
     return endpoint_.IsEndpoint(num_processed_frames, trailing_silence_frames,
                                 frame_shift_in_seconds);

sherpa-onnx/csrc/online-stream.cc

@@ -54,9 +54,20 @@ class OnlineStream::Impl {
 
   std::vector<Ort::Value> &GetStates() { return states_; }
 
+  int32_t &GetNumTrailingBlankFrames() { return num_trailing_blank_frames_; }
+
+  int32_t &GetWavSegment() { return segment_; }
+
+  int32_t &GetStartFrame() { return start_frame_; }
+
  private:
   FeatureExtractor feat_extractor_;
   int32_t num_processed_frames_ = 0;  // before subsampling
+  int32_t num_trailing_blank_frames_ = 0;  // after subsampling
+  /// ID of this segment
+  int32_t segment_ = 0;
+  /// Starting frame of this segment.
+  int32_t start_frame_ = 0;
   int32_t start_frame_index_ = 0;     // never reset
   OnlineTransducerDecoderResult result_;
   std::vector<Ort::Value> states_;
@@ -109,4 +120,15 @@ std::vector<Ort::Value> &OnlineStream::GetStates() {
   return impl_->GetStates();
 }
 
+int32_t &OnlineStream::GetNumTrailingBlankFrames() {
+  return impl_->GetNumTrailingBlankFrames();
+}
+
+int32_t &OnlineStream::GetWavSegment() {
+  return impl_->GetWavSegment();
+}
+
+int32_t &OnlineStream::GetStartFrame() {
+  return impl_->GetStartFrame();
+}
 }  // namespace sherpa_onnx

sherpa-onnx/csrc/online-stream.h

@@ -70,6 +70,18 @@ class OnlineStream {
 
   void SetStates(std::vector<Ort::Value> states);
   std::vector<Ort::Value> &GetStates();
+  // Used only for greedy search
+  //
+  // Get number of trailing blank frames decoded so far
+  //
+  // The returned reference is valid as long as this object is alive.
+  int32_t &GetNumTrailingBlankFrames();
+
+  // Return ID of this segment in Stream
+  int32_t &GetWavSegment();
+
+  // Return Starting frame of this segment.
+  int32_t &GetStartFrame();
 
  private:
   class Impl;