Tune engine vad config

engine/app/config.py

@@ -53,6 +53,10 @@ class Settings(BaseSettings):
     siliconflow_asr_model: str = Field(default="FunAudioLLM/SenseVoiceSmall", description="SiliconFlow ASR model")
     asr_interim_interval_ms: int = Field(default=500, description="Interval for interim ASR results in ms")
     asr_min_audio_ms: int = Field(default=300, description="Minimum audio duration before first ASR result")
+    asr_start_min_speech_ms: int = Field(
+        default=160,
+        description="Minimum continuous speech duration before ASR capture starts"
+    )
     asr_pre_speech_ms: int = Field(
         default=240,
         description="Audio context (ms) prepended before detected speech to avoid clipping first phoneme"