RealTimeSTT 0.3.99__tar.gz → 0.3.101__tar.gz

{realtimestt-0.3.99/realtimestt.egg-info → realtimestt-0.3.101}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: realtimestt
-Version: 0.3.99
+Version: 0.3.101
 Summary: A fast Voice Activity Detection and Transcription System
 Home-page: https://github.com/KoljaB/RealTimeSTT
 Author: Kolja Beigel
@@ -17,11 +17,10 @@ Requires-Dist: webrtcvad-wheels==2.0.14
 Requires-Dist: halo==0.0.31
 Requires-Dist: torch
 Requires-Dist: torchaudio
-Requires-Dist: scipy==1.15.1
+Requires-Dist: scipy==1.15.2
 Requires-Dist: openwakeword>=0.4.0
-Requires-Dist: websockets==14.2
+Requires-Dist: websockets==15.0.1
 Requires-Dist: websocket-client==1.8.0
-Requires-Dist: numpy<2.0.0
 Requires-Dist: soundfile==0.13.1
 Dynamic: author
 Dynamic: author-email
@@ -67,7 +66,7 @@ https://github.com/user-attachments/assets/797e6552-27cd-41b1-a7f3-e5cbc72094f5
 
 ### Updates
 
-Latest Version: v0.3.99
+Latest Version: v0.3.100
 
 See [release history](https://github.com/KoljaB/RealtimeSTT/releases).
 
@@ -553,6 +552,10 @@ When you initialize the `AudioToTextRecorder` class, you have various options to
 
 - **pre_recording_buffer_duration** (float, default=0.2): The time span, in seconds, during which audio is buffered prior to formal recording. This helps counterbalancing the latency inherent in speech activity detection, ensuring no initial audio is missed.
 
+- **on_vad_start**: A callable function triggered when the system has detected the start of voice activity presence.
+
+- **on_vad_stop**: A callable function triggered when the system has detected the stop of voice activity presence.
+
 - **on_vad_detect_start**: A callable function triggered when the system starts to listen for voice activity.
 
 - **on_vad_detect_stop**: A callable function triggered when the system stops to listen for voice activity.

{realtimestt-0.3.99 → realtimestt-0.3.101}/README.md

@@ -30,7 +30,7 @@ https://github.com/user-attachments/assets/797e6552-27cd-41b1-a7f3-e5cbc72094f5
 
 ### Updates
 
-Latest Version: v0.3.99
+Latest Version: v0.3.100
 
 See [release history](https://github.com/KoljaB/RealtimeSTT/releases).
 
@@ -516,6 +516,10 @@ When you initialize the `AudioToTextRecorder` class, you have various options to
 
 - **pre_recording_buffer_duration** (float, default=0.2): The time span, in seconds, during which audio is buffered prior to formal recording. This helps counterbalancing the latency inherent in speech activity detection, ensuring no initial audio is missed.
 
+- **on_vad_start**: A callable function triggered when the system has detected the start of voice activity presence.
+
+- **on_vad_stop**: A callable function triggered when the system has detected the stop of voice activity presence.
+
 - **on_vad_detect_start**: A callable function triggered when the system starts to listen for voice activity.
 
 - **on_vad_detect_stop**: A callable function triggered when the system stops to listen for voice activity.

{realtimestt-0.3.99 → realtimestt-0.3.101}/RealtimeSTT/audio_recorder.py

@@ -273,11 +273,13 @@ class AudioToTextRecorder:
                  pre_recording_buffer_duration: float = (
                      INIT_PRE_RECORDING_BUFFER_DURATION
                  ),
+                 on_vad_start=None,
+                 on_vad_stop=None,
                  on_vad_detect_start=None,
                  on_vad_detect_stop=None,
 
                  # Wake word parameters
-                 wakeword_backend: str = "pvporcupine",
+                 wakeword_backend: str = "",
                  openwakeword_model_paths: str = None,
                  openwakeword_inference_framework: str = "onnx",
                  wake_words: str = "",
@@ -422,10 +424,18 @@ class AudioToTextRecorder:
         - pre_recording_buffer_duration (float, default=0.2): Duration in
             seconds for the audio buffer to maintain pre-roll audio
             (compensates speech activity detection latency)
+        - on_vad_start (callable, default=None): Callback function to be called
+            when the system detected the start of voice activity presence.
+        - on_vad_stop (callable, default=None): Callback function to be called
+            when the system detected the stop (end) of voice activity presence.
         - on_vad_detect_start (callable, default=None): Callback function to
-            be called when the system listens for voice activity.
+            be called when the system listens for voice activity. This is not
+            called when VAD actually happens (use on_vad_start for this), but
+            when the system starts listening for it.
         - on_vad_detect_stop (callable, default=None): Callback function to be
-            called when the system stops listening for voice activity.
+            called when the system stops listening for voice activity. This is
+            not called when VAD actually stops (use on_vad_stop for this), but
+            when the system stops listening for it.
         - wakeword_backend (str, default="pvporcupine"): Specifies the backend
             library to use for wake word detection. Supported options include
             'pvporcupine' for using the Porcupine wake word engine or 'oww' for
@@ -550,6 +560,8 @@ class AudioToTextRecorder:
         self.on_recording_stop = on_recording_stop
         self.on_wakeword_detected = on_wakeword_detected
         self.on_wakeword_timeout = on_wakeword_timeout
+        self.on_vad_start = on_vad_start
+        self.on_vad_stop = on_vad_stop
         self.on_vad_detect_start = on_vad_detect_start
         self.on_vad_detect_stop = on_vad_detect_stop
         self.on_wakeword_detection_start = on_wakeword_detection_start
@@ -629,6 +641,7 @@ class AudioToTextRecorder:
         self.early_transcription_on_silence = early_transcription_on_silence
         self.use_extended_logging = use_extended_logging
         self.faster_whisper_vad_filter = faster_whisper_vad_filter
+        self.awaiting_speech_end = False
 
         # ----------------------------------------------------------------------------
         # Named logger configuration
@@ -759,7 +772,7 @@ class AudioToTextRecorder:
                           "transcription model initialized successfully")
 
         # Setup wake word detection
-        if wake_words or wakeword_backend in {'oww', 'openwakeword', 'openwakewords'}:
+        if wake_words or wakeword_backend in {'oww', 'openwakeword', 'openwakewords', 'pvp', 'pvporcupine'}:
             self.wakeword_backend = wakeword_backend
 
             self.wake_words_list = [
@@ -1427,41 +1440,15 @@ class AudioToTextRecorder:
             self.shutdown()
             raise  # Re-raise the exception after cleanup
 
-    def transcribe(self):
-        """
-        Transcribes audio captured by this class instance using the
-        `faster_whisper` model.
-
-        Automatically starts recording upon voice activity if not manually
-          started using `recorder.start()`.
-        Automatically stops recording upon voice deactivity if not manually
-          stopped with `recorder.stop()`.
-        Processes the recorded audio to generate transcription.
-
-        Args:
-            on_transcription_finished (callable, optional): Callback function
-              to be executed when transcription is ready.
-            If provided, transcription will be performed asynchronously,
-              and the callback will receive the transcription as its argument.
-              If omitted, the transcription will be performed synchronously,
-              and the result will be returned.
 
-        Returns (if no callback is set):
-            str: The transcription of the recorded audio.
-
-        Raises:
-            Exception: If there is an error during the transcription process.
-        """
-        self._set_state("transcribing")
-        audio_copy = copy.deepcopy(self.audio)
+    def perform_final_transcription(self, audio_bytes=None):
         start_time = 0
         with self.transcription_lock:
-            
             try:
                 if self.transcribe_count == 0:
                     logger.debug("Adding transcription request, no early transcription started")
                     start_time = time.time()  # Start timing
-                    self.parent_transcription_pipe.send((audio_copy, self.language))
+                    self.parent_transcription_pipe.send((audio_bytes, self.language))
                     self.transcribe_count += 1
 
                 while self.transcribe_count > 0:
@@ -1481,7 +1468,7 @@ class AudioToTextRecorder:
                     segments, info = result
                     self.detected_language = info.language if info.language_probability > 0 else None
                     self.detected_language_probability = info.language_probability
-                    self.last_transcription_bytes = copy.deepcopy(audio_copy)                    
+                    self.last_transcription_bytes = copy.deepcopy(audio_bytes)
                     self.last_transcription_bytes_b64 = base64.b64encode(self.last_transcription_bytes.tobytes()).decode('utf-8')
                     transcription = self._preprocess_output(segments)
                     end_time = time.time()  # End timing
@@ -1500,6 +1487,43 @@ class AudioToTextRecorder:
                 logger.error(f"Error during transcription: {str(e)}", exc_info=True)
                 raise e
 
+
+    def transcribe(self):
+        """
+        Transcribes audio captured by this class instance using the
+        `faster_whisper` model.
+
+        Automatically starts recording upon voice activity if not manually
+          started using `recorder.start()`.
+        Automatically stops recording upon voice deactivity if not manually
+          stopped with `recorder.stop()`.
+        Processes the recorded audio to generate transcription.
+
+        Args:
+            on_transcription_finished (callable, optional): Callback function
+              to be executed when transcription is ready.
+            If provided, transcription will be performed asynchronously,
+              and the callback will receive the transcription as its argument.
+              If omitted, the transcription will be performed synchronously,
+              and the result will be returned.
+
+        Returns (if no callback is set):
+            str: The transcription of the recorded audio.
+
+        Raises:
+            Exception: If there is an error during the transcription process.
+        """
+        audio_copy = copy.deepcopy(self.audio)
+        self._set_state("transcribing")
+        if self.on_transcription_start:
+            abort_value = self.on_transcription_start(audio_copy)
+            if not abort_value:
+                return self.perform_final_transcription(audio_copy)
+            return None
+        else:
+            return self.perform_final_transcription(audio_copy)
+
+
     def _process_wakeword(self, data):
         """
         Processes audio data to detect wake words.
@@ -1985,8 +2009,8 @@ class AudioToTextRecorder:
 
                         if self._is_voice_active():
 
-                            if self.on_vad_detect_start:
-                               self.on_vad_detect_start()
+                            if self.on_vad_start:
+                               self.on_vad_start()
 
                             if self.use_extended_logging:
                                 logger.debug('Debug: Voice activity detected')
@@ -2072,6 +2096,7 @@ class AudioToTextRecorder:
                                 (time.time() - self.recording_start_time > self.min_length_of_recording):
 
                                 self.speech_end_silence_start = time.time()
+                                self.awaiting_speech_end = True
 
                             if self.use_extended_logging:
                                 logger.debug('Debug: Checking early transcription conditions')
@@ -2092,6 +2117,7 @@ class AudioToTextRecorder:
                                     self.allowed_to_early_transcribe = False
 
                         else:
+                            self.awaiting_speech_end = False
                             if self.use_extended_logging:
                                 logger.debug('Debug: Handling speech detection')
                             if self.speech_end_silence_start:
@@ -2106,8 +2132,9 @@ class AudioToTextRecorder:
                         if self.speech_end_silence_start and time.time() - \
                                 self.speech_end_silence_start >= \
                                 self.post_speech_silence_duration:
-                            if self.on_vad_detect_stop:
-                                self.on_vad_detect_stop()
+
+                            if self.on_vad_stop:
+                                self.on_vad_stop()
 
                             if self.use_extended_logging:
                                 logger.debug('Debug: Formatting silence start time')
@@ -2139,6 +2166,8 @@ class AudioToTextRecorder:
                                     logger.debug('Debug: Setting failed_stop_attempt to True')
                                 failed_stop_attempt = True
 
+                            self.awaiting_speech_end = False
+
                 if self.use_extended_logging:
                     logger.debug('Debug: Checking if recording stopped')
                 if not self.is_recording and was_recording:
@@ -2194,6 +2223,8 @@ class AudioToTextRecorder:
             logger.debug('Debug: Exiting _recording_worker method')
 
 
+
+
     def _realtime_worker(self):
         """
         Performs real-time transcription if the feature is enabled.
@@ -2213,14 +2244,29 @@ class AudioToTextRecorder:
             if not self.enable_realtime_transcription:
                 return
 
-            # Continue running as long as the main process is active
+            # Track time of last transcription
+            last_transcription_time = time.time()
+
             while self.is_running:
 
-                # Check if the recording is active
                 if self.is_recording:
 
-                    # Sleep for the duration of the transcription resolution
-                    time.sleep(self.realtime_processing_pause)
+                    # MODIFIED SLEEP LOGIC:
+                    # Wait until realtime_processing_pause has elapsed,
+                    # but check often so we can respond to changes quickly.
+                    while (
+                        time.time() - last_transcription_time
+                    ) < self.realtime_processing_pause:
+                        time.sleep(0.001)
+                        if not self.is_running or not self.is_recording:
+                            break
+
+                    if self.awaiting_speech_end:
+                        time.sleep(0.001)
+                        continue
+
+                    # Update transcription time
+                    last_transcription_time = time.time()
 
                     # Convert the buffer frames to a NumPy array
                     audio_array = np.frombuffer(
@@ -2524,12 +2570,17 @@ class AudioToTextRecorder:
         logger.info(f"State changed from '{old_state}' to '{new_state}'")
 
         # Execute callbacks based on transitioning FROM a particular state
-        if old_state == "wakeword":
+        if old_state == "listening":
+            if self.on_vad_detect_stop:
+                self.on_vad_detect_stop()
+        elif old_state == "wakeword":
             if self.on_wakeword_detection_end:
                 self.on_wakeword_detection_end()
 
         # Execute callbacks based on transitioning TO a particular state
         if new_state == "listening":
+            if self.on_vad_detect_start:
+                self.on_vad_detect_start()
             self._set_spinner("speak now")
             if self.spinner and self.halo:
                 self.halo._interval = 250
@@ -2540,8 +2591,6 @@ class AudioToTextRecorder:
             if self.spinner and self.halo:
                 self.halo._interval = 500
         elif new_state == "transcribing":
-            if self.on_transcription_start:
-                self.on_transcription_start()
             self._set_spinner("transcribing")
             if self.spinner and self.halo:
                 self.halo._interval = 50

{realtimestt-0.3.99 → realtimestt-0.3.101}/RealtimeSTT/audio_recorder_client.py

@@ -257,10 +257,13 @@ class AudioToTextRecorderClient:
 
 
         if self.server_already_running:
-            self.set_parameter("language", self.language)
-            # self.set_parameter("model", self.model)
-
-
+            if not self.connection_established.wait(timeout=10):
+                print("Server connection not established within 10 seconds.")
+            else:
+                self.set_parameter("language", self.language)
+                print(f"Language set to {self.language}")
+                self.set_parameter("wake_word_activation_delay", self.wake_word_activation_delay)
+                print(f"Wake word activation delay set to {self.wake_word_activation_delay}")
 
     def text(self, on_transcription_finished=None):
         self.realtime_text = ""
@@ -490,11 +493,16 @@ class AudioToTextRecorderClient:
         print("STT server start command issued. Please wait a moment for it to initialize.", file=sys.stderr)
 
     def is_server_running(self):
-        parsed_url = urlparse(self.control_url)
-        host = parsed_url.hostname
-        port = parsed_url.port or 80
-        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
-            return s.connect_ex((host, port)) == 0
+        try:
+            # Attempt a proper WebSocket handshake to the control URL.
+            from websocket import create_connection
+            ws = create_connection(self.control_url, timeout=3)
+            ws.close()
+            return True
+        except Exception as e:
+            if self.debug_mode:
+                print(f"Server connectivity check failed: {e}")
+            return False
 
     def ensure_server_running(self):
         if not self.is_server_running():

{realtimestt-0.3.99 → realtimestt-0.3.101}/RealtimeSTT_server/stt_server.py

@@ -290,14 +290,12 @@ def text_detected(text, loop):
         print(f"\r[{timestamp}] {bcolors.OKCYAN}{text}{bcolors.ENDC}", flush=True, end='')
 
 def on_recording_start(loop):
-    # Send a message to the client indicating recording has started
     message = json.dumps({
         'type': 'recording_start'
     })
     asyncio.run_coroutine_threadsafe(audio_queue.put(message), loop)
 
 def on_recording_stop(loop):
-    # Send a message to the client indicating recording has stopped
     message = json.dumps({
         'type': 'recording_stop'
     })
@@ -316,28 +314,24 @@ def on_vad_detect_stop(loop):
     asyncio.run_coroutine_threadsafe(audio_queue.put(message), loop)
 
 def on_wakeword_detected(loop):
-    # Send a message to the client when wake word detection starts
     message = json.dumps({
         'type': 'wakeword_detected'
     })
     asyncio.run_coroutine_threadsafe(audio_queue.put(message), loop)
 
 def on_wakeword_detection_start(loop):
-    # Send a message to the client when wake word detection starts
     message = json.dumps({
         'type': 'wakeword_detection_start'
     })
     asyncio.run_coroutine_threadsafe(audio_queue.put(message), loop)
 
 def on_wakeword_detection_end(loop):
-    # Send a message to the client when wake word detection ends
     message = json.dumps({
         'type': 'wakeword_detection_end'
     })
     asyncio.run_coroutine_threadsafe(audio_queue.put(message), loop)
 
-def on_transcription_start(loop):
-    # Send a message to the client when transcription starts
+def on_transcription_start(_audio_bytes, loop):
     message = json.dumps({
         'type': 'transcription_start'
     })
@@ -460,8 +454,8 @@ def parse_arguments():
     parser.add_argument('--wake_word_timeout', type=float, default=5.0,
                         help='Maximum time in seconds that the system will wait for a wake word before timing out. After this timeout, the system stops listening for wake words until reactivated. Default is 5.0 seconds.')
 
-    parser.add_argument('--wake_word_activation_delay', type=float, default=20,
-                        help='The delay in seconds before the wake word detection is activated after the system starts listening. This prevents false positives during the start of a session. Default is 0.5 seconds.')
+    parser.add_argument('--wake_word_activation_delay', type=float, default=0,
+                        help='The delay in seconds before the wake word detection is activated after the system starts listening. This prevents false positives during the start of a session. Default is 0 seconds.')
 
     parser.add_argument('--wakeword_backend', type=str, default='none',
                         help='The backend used for wake word detection. You can specify different backends such as "default" or any custom implementations depending on your setup. Default is "pvporcupine".')
@@ -776,6 +770,7 @@ async def main_async():
         'use_main_model_for_realtime': args.use_main_model_for_realtime,
         'spinner': False,
         'use_microphone': False,
+
         'on_realtime_transcription_update': make_callback(loop, text_detected),
         'on_recording_start': make_callback(loop, on_recording_start),
         'on_recording_stop': make_callback(loop, on_recording_stop),

{realtimestt-0.3.99 → realtimestt-0.3.101/realtimestt.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: realtimestt
-Version: 0.3.99
+Version: 0.3.101
 Summary: A fast Voice Activity Detection and Transcription System
 Home-page: https://github.com/KoljaB/RealTimeSTT
 Author: Kolja Beigel
@@ -17,11 +17,10 @@ Requires-Dist: webrtcvad-wheels==2.0.14
 Requires-Dist: halo==0.0.31
 Requires-Dist: torch
 Requires-Dist: torchaudio
-Requires-Dist: scipy==1.15.1
+Requires-Dist: scipy==1.15.2
 Requires-Dist: openwakeword>=0.4.0
-Requires-Dist: websockets==14.2
+Requires-Dist: websockets==15.0.1
 Requires-Dist: websocket-client==1.8.0
-Requires-Dist: numpy<2.0.0
 Requires-Dist: soundfile==0.13.1
 Dynamic: author
 Dynamic: author-email
@@ -67,7 +66,7 @@ https://github.com/user-attachments/assets/797e6552-27cd-41b1-a7f3-e5cbc72094f5
 
 ### Updates
 
-Latest Version: v0.3.99
+Latest Version: v0.3.100
 
 See [release history](https://github.com/KoljaB/RealtimeSTT/releases).
 
@@ -553,6 +552,10 @@ When you initialize the `AudioToTextRecorder` class, you have various options to
 
 - **pre_recording_buffer_duration** (float, default=0.2): The time span, in seconds, during which audio is buffered prior to formal recording. This helps counterbalancing the latency inherent in speech activity detection, ensuring no initial audio is missed.
 
+- **on_vad_start**: A callable function triggered when the system has detected the start of voice activity presence.
+
+- **on_vad_stop**: A callable function triggered when the system has detected the stop of voice activity presence.
+
 - **on_vad_detect_start**: A callable function triggered when the system starts to listen for voice activity.
 
 - **on_vad_detect_stop**: A callable function triggered when the system stops to listen for voice activity.

{realtimestt-0.3.99 → realtimestt-0.3.101}/realtimestt.egg-info/requires.txt

@@ -5,9 +5,8 @@ webrtcvad-wheels==2.0.14
 halo==0.0.31
 torch
 torchaudio
-scipy==1.15.1
+scipy==1.15.2
 openwakeword>=0.4.0
-websockets==14.2
+websockets==15.0.1
 websocket-client==1.8.0
-numpy<2.0.0
 soundfile==0.13.1

{realtimestt-0.3.99 → realtimestt-0.3.101}/requirements.txt

@@ -5,9 +5,8 @@ webrtcvad-wheels==2.0.14
 halo==0.0.31
 torch
 torchaudio
-scipy==1.15.1
+scipy==1.15.2
 openwakeword>=0.4.0
-websockets==14.2
+websockets==15.0.1
 websocket-client==1.8.0
-numpy<2.0.0
 soundfile==0.13.1

{realtimestt-0.3.99 → realtimestt-0.3.101}/setup.py

@@ -14,7 +14,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
 
 setuptools.setup(
     name="realtimestt",
-    version="0.3.99",
+    version="0.3.101",
     author="Kolja Beigel",
     author_email="kolja.beigel@web.de",
     description="A fast Voice Activity Detection and Transcription System",