Move to 3-value endpointer config

python/example/test_ep.py

@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel, EndpointerMode
+
+# You can set log level to -1 to disable debug messages
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+
+# You can also init model by name or with a folder path
+# model = Model(model_name="vosk-model-en-us-0.21")
+# model = Model("models/en")
+
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
+rec.SetPartialWords(True)
+rec.SetEndpointerMode(EndpointerMode.VERY_LONG)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())
+
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+rec.SetEndpointerDelays(300, 500, 2000)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())

python/setup.py

@@ -45,7 +45,7 @@ with open("README.md", "rb") as fh:
 
 setuptools.setup(
     name="vosk",
-    version="0.3.45",
+    version="0.3.46",
     author="Alpha Cephei Inc",
     author_email="contact@alphacephei.com",
     description="Offline open source speech recognition API based on Kaldi and Vosk",

python/vosk/__init__.py

@@ -142,10 +142,17 @@ class SpkModel:
     def __del__(self):
         _c.vosk_spk_model_free(self._handle)
 
-class EpMode(enum.Enum):
+class EndpointerMode(enum.Enum):
     DEFAULT = 0
-    SHORT = 1
-    LONG = 2
+    SINGLE_WORD = 1
+    VERY_SHORT = 2
+    SHORT = 3
+    STANDARD = 4
+    STANDARD_5 = 5
+    STANDARD_180 = 6
+    LONG = 7
+    VERY_LONG = 8
+    VERY_LONG_180 = 9
 
 class KaldiRecognizer:
 
@@ -179,8 +186,11 @@ class KaldiRecognizer:
     def SetNLSML(self, enable_nlsml):
         _c.vosk_recognizer_set_nlsml(self._handle, 1 if enable_nlsml else 0)
 
-    def SetEpMode(self, mode):
-        _c.vosk_recognizer_set_ep_mode(self._handle, mode.value)
+    def SetEndpointerMode(self, mode):
+        _c.vosk_recognizer_set_endpointer_mode(self._handle, mode.value)
+
+    def SetEndpointerDelays(self, sct, nit, t):
+        _c.vosk_recognizer_set_endpointer_delays(self._handle, sct, nit, t)
 
     def SetSpkModel(self, spk_model):
         _c.vosk_recognizer_set_spk_model(self._handle, spk_model._handle)

src/recognizer.cc

@@ -220,26 +220,121 @@ void Recognizer::SetNLSML(bool nlsml)
     nlsml_ = nlsml;
 }
 
-void Recognizer::SetEpMode(int mode)
+void Recognizer::SetEndpointerMode(int mode)
 {
+    float rule1, rule2, rule3, rule4, rule5;
+    if (mode == 0) {
+        endpoint_config_ = model_->endpoint_config_;
+        return;
+    }
+
     float scale = 1.0;
     switch(mode) {
-        case 1:
-           scale = 0.75;
+        case 1: // SINGLE_WORD
+           rule2 = 0.1;
+           rule3 = 0.15;
+           rule4 = 0.2;
+           rule1 = 4.0;
+           rule5 = 7.0;
            break;
-        case 2:
-           scale = 1.50;
+        case 2: // VERY_SHORT
+           rule2 = 0.3;
+           rule3 = 0.3;
+           rule4 = 0.6;
+           rule1 = 3.0;
+           rule5 = 5.0;
+           break;
+        case 3: // SHORT
+           rule2 = 0.3;
+           rule3 = 0.4;
+           rule4 = 0.6;
+           rule1 = 3.0;
+           rule5 = 5.0;
+           break;
+        case 4: // STANDARD
+           rule2 = 0.8;
+           rule3 = 0.9;
+           rule4 = 1.0;
+           rule1 = 3.0;
+           rule5 = 7.0;
+           break;
+        case 5: // STANDARD_5
+           rule2 = 0.8;
+           rule3 = 0.9;
+           rule4 = 1.5;
+           rule1 = 5.0;
+           rule5 = 7.0;
+           break;
+        case 6: // STANDARD_180
+           rule2 = 0.8;
+           rule3 = 0.9;
+           rule4 = 1.5;
+           rule1 = 3.0;
+           rule5 = 180.0;
+           break;
+        case 7: // LONG
+           rule2 = 1.0;
+           rule3 = 1.2;
+           rule4 = 2.0;
+           rule1 = 4.0;
+           rule5 = 10.0;
+           break;
+        case 8: // VERY_LONG
+           rule2 = 2.0;
+           rule3 = 2.5;
+           rule4 = 3.0;
+           rule1 = 4.0;
+           rule5 = 15.0;
+           break;
+        case 9: // VERY_LONG_180
+           rule2 = 2.0;
+           rule3 = 2.5;
+           rule4 = 3.0;
+           rule1 = 4.0;
+           rule5 = 180.0;
+           break;
+        default: // STANDARD
+           rule2 = 0.8;
+           rule3 = 0.9;
+           rule4 = 1.0;
+           rule1 = 5.0;
+           rule5 = 7.0;
            break;
-        default:
-           scale = 4.0;
     }
-    KALDI_LOG << "Endpointer Scale " << scale;
+    KALDI_LOG << "Updating endpointer timeouts to " << rule1 << "," << rule2 << "," << rule3 << "," << rule4 << "," << rule5;
     endpoint_config_ = model_->endpoint_config_;
-    endpoint_config_.rule2.min_trailing_silence *= scale;
-    endpoint_config_.rule3.min_trailing_silence *= scale;
-    endpoint_config_.rule4.min_trailing_silence *= scale;
+    endpoint_config_.rule1.min_trailing_silence = rule1;
+    endpoint_config_.rule2.min_trailing_silence = rule2;
+    endpoint_config_.rule3.min_trailing_silence = rule3;
+    endpoint_config_.rule4.min_trailing_silence = rule4;
+    endpoint_config_.rule5.min_utterance_length = rule5;
 }
 
+void Recognizer::SetEndpointerDelays(int sct, int nit, int t)
+{
+    float rule1, rule2, rule3, rule4, rule5;
+
+    rule2 = sct / 1000.0;
+    if (sct < 500) {
+        rule3 = sct / 1000.0 * 1.5;
+        rule4 = sct / 1000.0 * 2;
+    } else  {
+        rule3 = sct / 1000.0 + 0.5;
+        rule4 = sct / 1000.0 + 1.0;
+    }
+    rule1 = nit / 1000.0;
+    rule5 = t / 1000.0;
+
+    KALDI_LOG << "Updating endpointer timeouts to " << rule1 << "," << rule2 << "," << rule3 << "," << rule4 << "," << rule5;
+    endpoint_config_ = model_->endpoint_config_;
+    endpoint_config_.rule1.min_trailing_silence = rule1;
+    endpoint_config_.rule2.min_trailing_silence = rule2;
+    endpoint_config_.rule3.min_trailing_silence = rule3;
+    endpoint_config_.rule4.min_trailing_silence = rule4;
+    endpoint_config_.rule5.min_utterance_length = rule5;
+}
+
+
 void Recognizer::SetSpkModel(SpkModel *spk_model)
 {
     if (state_ == RECOGNIZER_RUNNING) {

src/recognizer.h

@@ -52,7 +52,8 @@ class Recognizer {
         void SetWords(bool words);
         void SetPartialWords(bool partial_words);
         void SetNLSML(bool nlsml);
-        void SetEpMode(int mode);
+        void SetEndpointerMode(int mode);
+        void SetEndpointerDelays(int sct, int nit, int t);
         bool AcceptWaveform(const char *data, int len);
         bool AcceptWaveform(const short *sdata, int len);
         bool AcceptWaveform(const float *fdata, int len);

src/vosk_api.cc

@@ -129,12 +129,20 @@ void vosk_recognizer_set_grm(VoskRecognizer *recognizer, char const *grammar)
     ((Recognizer *)recognizer)->SetGrm(grammar);
 }
 
-void vosk_recognizer_set_ep_mode(VoskRecognizer *recognizer, VoskEpMode mode)
+void vosk_recognizer_set_endpointer_mode(VoskRecognizer *recognizer, VoskEndpointerMode mode)
 {
     if (recognizer == nullptr) {
        return;
     }
-    ((Recognizer *)recognizer)->SetEpMode(mode);
+    ((Recognizer *)recognizer)->SetEndpointerMode(mode);
+}
+
+void vosk_recognizer_set_endpointer_delays(VoskRecognizer *recognizer, int sct, int nit, int t)
+{
+    if (recognizer == nullptr) {
+       return;
+    }
+    ((Recognizer *)recognizer)->SetEndpointerDelays(sct, nit, t);
 }
 
 int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length)

src/vosk_api.h

@@ -217,18 +217,35 @@ void vosk_recognizer_set_partial_words(VoskRecognizer *recognizer, int partial_w
  */
 void vosk_recognizer_set_nlsml(VoskRecognizer *recognizer, int nlsml);
 
-typedef enum VoskEpMode {
-    VOSK_EP_ANSWER_DEFAULT = 0,
-    VOSK_EP_ANSWER_SHORT = 1,
-    VOSK_EP_ANSWER_LONG = 2,
-} VoskEpMode;
+typedef enum VoskEndpointerMode {
+    VOSK_EP_DEFAULT = 0,
+    VOSK_EP_SINGLE_WORD = 1,
+    VOSK_EP_VERY_SHORT = 2,
+    VOSK_EP_SHORT = 3,
+    VOSK_EP_STANDARD = 4,
+    VOSK_EP_STANDARD_5 = 5,
+    VOSK_EP_STANDARD_180 = 6,
+    VOSK_EP_LONG = 7,
+    VOSK_EP_VERY_LONG = 8,
+    VOSK_EP_VERY_LONG_180 = 9
+} VoskEndpointerMode;
 
 /**
  * Set endpointer scaling factor
  *
  * @param mode - Endpointer mode
  **/
-void vosk_recognizer_set_ep_mode(VoskRecognizer *recognizer,  VoskEpMode mode);
+void vosk_recognizer_set_endpointer_mode(VoskRecognizer *recognizer,  VoskEndpointerMode mode);
+
+/**
+ * Set endpointer delays
+ *
+ * @param sct   speech complete timeout
+ * @param sint  speech incomplete timeout
+ * @param nit   no input timeout
+ * @param t     recognition timeout
+ **/
+void vosk_recognizer_set_endpointer_delays(VoskRecognizer *recognizer, int sct, int nit, int t);
 
 /** Accept voice data
  *