From b0262cfe72e6f4416629314408b5816155137923 Mon Sep 17 00:00:00 2001
From: Sakura-RanChen <1908198662@qq.com>
Date: Tue, 29 Apr 2025 17:14:53 +0800
Subject: [PATCH] =?UTF-8?q?update:=E4=BC=98=E5=8C=96=E5=90=9E=E9=9F=B3?=
 =?UTF-8?q?=E9=97=AE=E9=A2=98=EF=BC=8C=E8=A7=A3=E5=86=B3=E6=92=AD=E6=94=BE?=
 =?UTF-8?q?=E9=95=BF=E9=9F=B3=E9=A2=91=E6=97=B6=E7=9A=84=E9=94=99=E8=AF=AF?=
 =?UTF-8?q?=E9=80=80=E5=87=BA?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 main/xiaozhi-server/core/connection.py        | 10 +++++--
 .../core/handle/sendAudioHandle.py            | 28 +++++++++++++------
 2 files changed, 26 insertions(+), 12 deletions(-)

diff --git a/main/xiaozhi-server/core/connection.py b/main/xiaozhi-server/core/connection.py
index 3cf6b018..c9e984f2 100644
--- a/main/xiaozhi-server/core/connection.py
+++ b/main/xiaozhi-server/core/connection.py
@@ -202,12 +202,16 @@ 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 ec29a9fd..1d4da583 100644
--- a/main/xiaozhi-server/core/handle/sendAudioHandle.py
+++ b/main/xiaozhi-server/core/handle/sendAudioHandle.py
@@ -51,8 +51,8 @@ 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)
 
@@ -64,22 +64,32 @@ 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()