vac model is loaded in TranscriptionEngine, and by default

whisperlivekit/audio_processor.py

@@ -66,10 +66,12 @@ class AudioProcessor:
         self.asr = models.asr
         self.tokenizer = models.tokenizer
         self.diarization = models.diarization
-        import torch
-        model, _ = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")
-        self.vac = FixedVADIterator(model)
-        self.vac.reset_states()
+        self.vac_model = models.vac_model
+        if self.args.vac:
+            self.vac = FixedVADIterator(models.vac_model)
+        else:
+            self.vac = None
+            
         self.ffmpeg_manager = FFmpegManager(
             sample_rate=self.sample_rate,
             channels=self.channels
@@ -218,11 +220,14 @@ class AudioProcessor:
                     # Process audio chunk
                     pcm_array = self.convert_pcm_to_float(self.pcm_buffer[:self.max_bytes_per_sec])
                     self.pcm_buffer = self.pcm_buffer[self.max_bytes_per_sec:]
-                    res = self.vac(pcm_array)
-
+                    
+                    res = None
                     end_of_audio = False
                     silence_buffer = None
                     
+                    if self.args.vac:
+                        res = self.vac(pcm_array)
+                    
                     if self.silence:
                         print('NO AUDIO')
                     

whisperlivekit/core.py

@@ -34,7 +34,7 @@ class TranscriptionEngine:
             "lan": "auto",
             "task": "transcribe",
             "backend": "faster-whisper",
-            "vac": False,
+            "vac": True,
             "vac_chunk_size": 0.04,
             "log_level": "DEBUG",
             "ssl_certfile": None,
@@ -82,6 +82,11 @@ class TranscriptionEngine:
         self.asr = None
         self.tokenizer = None
         self.diarization = None
+        self.vac_model = None
+        
+        if self.args.vac:
+            import torch
+            self.vac_model, _ = torch.hub.load(repo_or_dir="snakers4/silero-vad", model="silero_vad")            
         
         if self.args.transcription:
             if self.args.backend == "simulstreaming": 
@@ -131,7 +136,7 @@ def online_factory(args, asr, tokenizer, logfile=sys.stderr):
             logfile=logfile,
         )
         # warmup_online(online, args.warmup_file)
-    elif args.vac:
+    elif False: #args.vac: #vac is now handled in audio_processor
         online = VACOnlineASRProcessor(
             args.min_chunk_size,
             asr,

whisperlivekit/parse_args.py

@@ -113,8 +113,8 @@ def parse_args():
     )
     parser.add_argument(
         "--vac",
-        action="store_true",
-        default=False,
+        # action="store_true",
+        default=True,
         help="Use VAC = voice activity controller. Recommended. Requires torch.",
     )
     parser.add_argument(