adapter to run client

transcribe/client.py

@@ -133,9 +133,9 @@ class Client:
         """
         message = json.loads(message)
 
-        if self.uid != message.get("uid"):
-            print("[ERROR]: invalid client uid")
-            return
+        # if self.uid != message.get("uid"):
+        #     print("[ERROR]: invalid client uid")
+        #     return
 
         if "status" in message.keys():
             self.handle_status_messages(message)

transcribe/transcription.py

@@ -174,7 +174,8 @@ class TranscriptionServer:
         frame_data = websocket.recv()
         if frame_data == b"END_OF_AUDIO":
             return False
-        return np.frombuffer(frame_data, dtype=np.int16).astype(np.float32) / 32768.0
+        # return np.frombuffer(frame_data, dtype=np.int16).astype(np.float32) / 32768.0
+        return np.frombuffer(frame_data, dtype=np.float32)
 
 
     def handle_new_connection(self, websocket):
@@ -184,9 +185,11 @@ class TranscriptionServer:
        
         try:
             logging.info("New client connected")
-            # options = websocket.recv()
-            # options = json.loads(options)
-            options = {"language": from_lang, "uid": str(uuid1())}
+            options = websocket.recv()
+            try:
+                options = json.loads(options)
+            except Exception as e:
+                options = {"language": from_lang, "uid": str(uuid1())}
             if self.client_manager is None:
                 max_clients = options.get('max_clients', 4)
                 max_connection_time = options.get('max_connection_time', 600)

transcribe/whisper_llm_serve.py

@@ -117,7 +117,6 @@ class PywhisperInference:
     llm_model = None
     vad_model = None
  
-
     @classmethod
     def initializer(cls, event:mp.Event, warmup=True):
         models_dir = config.MODEL_DIR.as_posix()
@@ -265,9 +264,10 @@ class PyWhiperCppServe(ServeClientBase):
     def translate_text(self, text):
         """
         translate the text to dst lang"""
-        translate_fut = self._pool.submit(
-            PywhisperInference.translate, text, self.language, self.dst_lang)
-        return  translate_fut.result()
+        return "sample english"
+        # translate_fut = self._pool.submit(
+        #     PywhisperInference.translate, text, self.language, self.dst_lang)
+        # return  translate_fut.result()
      
     def _segments_split(self, segments, audio_buffer: np.ndarray):
         """根据左边第一个标点符号来将序列拆分成 观察段 和 剩余部分"""
@@ -292,7 +292,7 @@ class PyWhiperCppServe(ServeClientBase):
                 rest_buffer_duration = (len(audio_buffer) - seg_index) / self.sample_rate
                 # is_end = any(i in seg.text for i  in config.SENTENCE_END_MARKERS)
                 right_watch_sequences = segments[min(idx+1, len(segments)):]
-                if rest_buffer_duration >= 1:
+                if rest_buffer_duration >= 1.5:
                     left_watch_idx = seg_index
                 break
         return left_watch_idx, left_watch_sequences, right_watch_sequences, is_end