bump version v0.0.11

Fix: error messages.

whisper_live/__version__.py

@@ -1 +1 @@
-__version__="0.0.10"
+__version__="0.0.11"

whisper_live/client.py

@@ -412,7 +412,7 @@ class Client:
             for _ in range(0, int(self.rate / self.chunk * self.record_seconds)):
                 if not self.recording:
                     break
-                data = self.stream.read(self.chunk)
+                data = self.stream.read(self.chunk, exception_on_overflow = False)
                 self.frames += data
 
                 audio_array = Client.bytes_to_float_array(data)

whisper_live/server.py

@@ -102,7 +102,9 @@ class TranscriptionServer:
             language=options["language"],
             task=options["task"],
             client_uid=options["uid"],
-            model_size=options["model_size"]
+            model_size=options["model_size"],
+            initial_prompt=options.get("initial_prompt"),
+            vad_parameters=options.get("vad_parameters")
         )
         
         self.clients[websocket] = client
@@ -118,7 +120,7 @@ class TranscriptionServer:
                 elapsed_time = time.time() - self.clients_start_time[websocket]
                 if elapsed_time >= self.max_connection_time:
                     self.clients[websocket].disconnect()
-                    logging.warning(f"{self.clients[websocket]} Client disconnected due to overtime.")
+                    logging.warning(f"Client with uid '{self.clients[websocket].client_uid}' disconnected due to overtime.")
                     self.clients[websocket].cleanup()
                     self.clients.pop(websocket)
                     self.clients_start_time.pop(websocket)
@@ -127,13 +129,11 @@ class TranscriptionServer:
                     break
 
             except Exception as e:
-                logging.error(e)
+                logging.info(f"[ERROR]: Client with uid '{self.clients[websocket].client_uid}' Disconnected.")
                 if self.clients[websocket].model_size is not None:
                     self.clients[websocket].cleanup()
                 self.clients.pop(websocket)
                 self.clients_start_time.pop(websocket)
-                logging.info("Connection Closed.")
-                logging.info(self.clients)
                 del websocket
                 break
 
@@ -190,7 +190,9 @@ class ServeClient:
         multilingual=False,
         language=None,
         client_uid=None,
-        model_size="small"
+        model_size="small",
+        initial_prompt=None,
+        vad_parameters=None
         ):
         """
         Initialize a ServeClient instance.
@@ -218,6 +220,8 @@ class ServeClient:
         self.language = language if self.multilingual else "en"
         self.task = task
         self.websocket = websocket
+        self.initial_prompt = initial_prompt
+        self.vad_parameters = vad_parameters or {"threshold": 0.5}
         
         device = "cuda" if torch.cuda.is_available() else "cpu"
         
@@ -352,11 +356,11 @@ class ServeClient:
                 # whisper transcribe with prompt
                 result, info = self.transcriber.transcribe(
                     input_sample, 
-                    initial_prompt=None,
+                    initial_prompt=self.initial_prompt,
                     language=self.language,
                     task=self.task,
                     vad_filter=True,
-                    vad_parameters={"threshold": 0.5}
+                    vad_parameters=self.vad_parameters
                 )
 
                 if self.language is None:
@@ -401,12 +405,20 @@ class ServeClient:
                         })
                     )
                 except Exception as e:
-                    logging.error(f"[ERROR]: {e}")
+                    logging.error(f"[ERROR]: Failed to send message to client: {e}")
 
             except Exception as e:
-                logging.error(f"[ERROR]: {e}")
+                logging.error(f"[ERROR]: Failed to transcribe audio chunk: {e}")
                 time.sleep(0.01)
     
+    def format_segment(self, start, end, text):
+        """Helper function to format a segment with string timestamps."""
+        return {
+            'start': "{:.3f}".format(start),
+            'end': "{:.3f}".format(end),
+            'text': text
+        }
+
     def update_segments(self, segments, duration):
         """
         Processes the segments from whisper. Appends all the segments to the list
@@ -437,22 +449,16 @@ class ServeClient:
                 text_ = s.text
                 self.text.append(text_)
                 start, end = self.timestamp_offset + s.start, self.timestamp_offset + min(duration, s.end)
-                self.transcript.append(
+                self.transcript.append(self.format_segment(start, end, text_))
-                    {
-                        'start': start,
-                        'end': end,
-                        'text': text_
-                    }
-                )
                 
                 offset = min(duration, s.end)
 
         self.current_out += segments[-1].text
-        last_segment = {
+        last_segment = self.format_segment(
-            'start': self.timestamp_offset + segments[-1].start,
+            self.timestamp_offset + segments[-1].start,
-            'end': self.timestamp_offset + min(duration, segments[-1].end),
+            self.timestamp_offset + min(duration, segments[-1].end),
-            'text': self.current_out
+            self.current_out
-        }
+        )
         
         # if same incomplete segment is seen multiple times then update the offset
         # and append the segment to the list
@@ -464,13 +470,11 @@ class ServeClient:
         if self.same_output_threshold > 5:
             if not len(self.text) or self.text[-1].strip().lower()!=self.current_out.strip().lower():          
                 self.text.append(self.current_out)
-                self.transcript.append(
+                self.transcript.append(self.format_segment(
-                    {
+                    self.timestamp_offset,
-                        'start': self.timestamp_offset,
+                    self.timestamp_offset + duration,
-                        'end': self.timestamp_offset + duration,
+                    self.current_out
-                        'text': self.current_out
+                ))
-                    }
-                )
             self.current_out = ''
             offset = duration
             self.same_output_threshold = 0
@@ -512,4 +516,3 @@ class ServeClient:
         """
         logging.info("Cleaning up.")
         self.exit = True
-        self.transcriber.destroy()

whisper_live/transcriber.py

@@ -934,9 +934,6 @@ class WhisperModel:
             )
         ]
 
-    def destroy(self):
-        del self.model
-
 
 def restore_speech_timestamps(
     segments: Iterable[Segment],