Handle overflow errors in audio capture by adding `exception_on_overflow=False` and skip processing when no data is available.

src/voice_dialogue/core/launcher.py

@@ -63,7 +63,6 @@ def launch_system(
 
     # 音频采集
     enable_echo_cancellation = not disable_echo_cancellation
-    logger.info(f"Echo Cancellation: {enable_echo_cancellation}")
     audio_frame_probe = AudioCapture(
         audio_frames_queue=audio_frames_queue,
         enable_echo_cancellation=enable_echo_cancellation
@@ -73,7 +72,6 @@ def launch_system(
 
     # 语音状态监测
     enable_vad = disable_echo_cancellation
-    logger.info(f"VAD: {enable_vad}")
     user_voice_checker = SpeechStateMonitor(
         audio_frame_queue=audio_frames_queue,
         user_voice_queue=user_voice_queue,

src/voice_dialogue/services/audio/capture.py

@@ -92,7 +92,9 @@ class AudioCapture(BaseThread):
 
         try:
             while not self.is_exited:
-                data = stream.read(chunk)
+                data = stream.read(chunk, exception_on_overflow=False)
+                if data is None:
+                    continue
 
                 if self.is_paused:
                     continue