Add emotion, event of SenseVoice. (#1257)

* Add emotion, event of SenseVoice.

* Fix tokens size check and update java api.

https://github.com/k2-fsa/sherpa-onnx/pull/1257

sherpa-onnx/csrc/offline-recognizer-sense-voice-impl.h

@@ -52,6 +52,13 @@ static OfflineRecognitionResult ConvertSenseVoiceResult(
 
   r.words = std::move(src.words);
 
+  // parse lang, emotion and event from tokens.
+  if (src.tokens.size() >= 3) {
+    r.lang = sym_table[src.tokens[0]];
+    r.emotion = sym_table[src.tokens[1]];
+    r.event = sym_table[src.tokens[2]];
+  }
+
   return r;
 }
 

sherpa-onnx/csrc/offline-stream.cc

@@ -304,6 +304,19 @@ const OfflineRecognitionResult &OfflineStream::GetResult() const {
 std::string OfflineRecognitionResult::AsJsonString() const {
   std::ostringstream os;
   os << "{";
+
+  os << "\"lang\""
+     << ": ";
+  os << std::quoted(lang) << ", ";
+
+  os << "\"emotion\""
+     << ": ";
+  os << std::quoted(emotion) << ", ";
+
+  os << "\"event\""
+     << ": ";
+  os << std::quoted(event) << ", ";
+
   os << "\"text\""
      << ": ";
   os << std::quoted(text) << ", ";

sherpa-onnx/csrc/offline-stream.h

@@ -28,6 +28,12 @@ struct OfflineRecognitionResult {
 
   std::string lang;
 
+  // emotion target of the audio.
+  std::string emotion;
+
+  // event target of the audio.
+  std::string event;
+
     /// timestamps.size() == tokens.size()
   /// timestamps[i] records the time in seconds when tokens[i] is decoded.
   std::vector<float> timestamps;