diff --git a/main/xiaozhi-server/core/connection.py b/main/xiaozhi-server/core/connection.py
index 8f406ed1..fbd3569e 100644
--- a/main/xiaozhi-server/core/connection.py
+++ b/main/xiaozhi-server/core/connection.py
@@ -203,12 +203,16 @@ class ConnectionHandler:
         finally:
             await self.close(ws)
 
+    async def reset_timeout(self):
+        """重置超时计时器"""
+        if self.timeout_task:
+            self.timeout_task.cancel()
+        self.timeout_task = asyncio.create_task(self._check_timeout())
+
     async def _route_message(self, message):
         """消息路由"""
         # 重置超时计时器
-        if self.timeout_task:
-            self.timeout_task.cancel()
-            self.timeout_task = asyncio.create_task(self._check_timeout())
+        await self.reset_timeout()
 
         if isinstance(message, str):
             await handleTextMessage(self, message)
diff --git a/main/xiaozhi-server/core/handle/sendAudioHandle.py b/main/xiaozhi-server/core/handle/sendAudioHandle.py
index 3387a133..b9148238 100644
--- a/main/xiaozhi-server/core/handle/sendAudioHandle.py
+++ b/main/xiaozhi-server/core/handle/sendAudioHandle.py
@@ -52,8 +52,8 @@ async def sendAudioMessage(conn, audios, text, text_index=0):
         logger.bind(tag=TAG).info(f"发送第一段语音: {text}")
     await send_tts_message(conn, "sentence_start", text)
 
-    # 播放音频
-    await sendAudio(conn, audios)
+    is_first_audio = (text_index == conn.tts_first_text_index)
+    await sendAudio(conn, audios, pre_buffer=is_first_audio)
 
     await send_tts_message(conn, "sentence_end", text)
 
@@ -65,22 +65,32 @@ async def sendAudioMessage(conn, audios, text, text_index=0):
 
 
 # 播放音频
-async def sendAudio(conn, audios):
+async def sendAudio(conn, audios, pre_buffer=True):
     # 流控参数优化
     frame_duration = 60  # 帧时长（毫秒），匹配 Opus 编码
     start_time = time.perf_counter()
     play_position = 0
+    last_reset_time = time.perf_counter()  # 记录最后的重置时间
 
-    # 预缓冲：发送前 3 帧
-    pre_buffer = min(3, len(audios))
-    for i in range(pre_buffer):
-        await conn.websocket.send(audios[i])
+    # 仅当第一句话时执行预缓冲
+    if pre_buffer:
+        pre_buffer_frames = min(3, len(audios))
+        for i in range(pre_buffer_frames):
+            await conn.websocket.send(audios[i])
+        remaining_audios = audios[pre_buffer_frames:]
+    else:
+        remaining_audios = audios
 
-    # 正常播放剩余帧
-    for opus_packet in audios[pre_buffer:]:
+    # 播放剩余音频帧
+    for opus_packet in remaining_audios:
         if conn.client_abort:
             return
 
+        # 每分钟重置一次计时器
+        if time.perf_counter() - last_reset_time > 60:
+            await conn.reset_timeout()
+            last_reset_time = time.perf_counter()
+
         # 计算预期发送时间
         expected_time = start_time + (play_position / 1000)
         current_time = time.perf_counter()