diff --git a/app/endpoints/pipeline_endpoint.py b/app/endpoints/pipeline_endpoint.py
index 808ca10..943732e 100644
--- a/app/endpoints/pipeline_endpoint.py
+++ b/app/endpoints/pipeline_endpoint.py
@@ -17,7 +17,7 @@ from common.memory import *
 import traceback
 import sys
 
-timeout_duration = 8.0 
+timeout_duration = 2.0 
 
 messages_router = APIRouter()
 
@@ -454,6 +454,8 @@ async def handle_text_async(request: Request,token_id,app_id, wxid,msg_data,from
 
             return
         
+        request.app.state.message_lock[app_id]= asyncio.Lock()
+
         # 创建并启动任务协程，将参数传递给 ai_chat_text 函数
         task = asyncio.create_task(
             ai_chat_text_async(request,token_id, app_id, wxid, msg_data, msg_content)
@@ -469,11 +471,12 @@ async def handle_text_async(request: Request,token_id,app_id, wxid,msg_data,from
         # 取消定时器
         timeout_timer.cancel()
         
-message_lock = asyncio.Lock()
+#message_lock = asyncio.Lock()
 
 async def check_timeout_async(task: asyncio.Task, request: Request,token_id, wxid, app_id, callback_to_user):
     await asyncio.sleep(timeout_duration)  # 等待超时时间
     if not task.done():
+        message_lock=request.app.state.message_lock[app_id]
         async with message_lock:
             print(f"任务运行时间超过{timeout_duration}秒，token_id={token_id}, app_id={app_id}, callback_to_user={callback_to_user}")
             wx_config = await request.app.state.gewe_service.get_wxchat_config_from_cache_async(wxid)
@@ -588,6 +591,7 @@ async def ai_chat_text_async(request: Request,token_id, app_id, wxid, msg_data,
         await request.app.state.gewe_service.save_session_messages_to_cache_async(hash_key, {"role": "assistant", "content": reply_content})
         return    
 
+    message_lock=request.app.state.message_lock[app_id]
     # 昵称替换
     replacements = {
         '{昵称}': wxid_nickname,
@@ -598,6 +602,7 @@ async def ai_chat_text_async(request: Request,token_id, app_id, wxid, msg_data,
     # 判断图片url
     img_urls,reply_content=extract_and_replace_image_urls(reply_content)
     if img_urls:
+        
         #await request.app.state.gewe_service.post_text_async(token_id, app_id, callback_to_user, reply_content)
         async with message_lock:
             await ai_post_text_split_async_async(request, token_id, app_id, callback_to_user, reply_content)
@@ -872,6 +877,8 @@ async def handle_voice_async(request: Request,token_id,app_id, wxid,msg_data,fro
     msg_content=msg_data["Content"]["string"]
     msg_id=msg_data["MsgId"]
 
+    request.app.state.message_lock[app_id]=asyncio.Lock()
+
     config=await request.app.state.gewe_service.get_wxchat_config_from_cache_async(wxid)
     validated_config = AgentConfig.model_validate(config)
     if not validated_config.privateGroupChatEnabled:
@@ -918,6 +925,7 @@ async def handle_voice_async(request: Request,token_id,app_id, wxid,msg_data,fro
         # 判断图片url
         img_urls,reply_content=extract_and_replace_image_urls(reply_content)
         if img_urls:
+            message_lock=request.app.state.message_lock[app_id]
             #await request.app.state.gewe_service.post_text_async(token_id, app_id, callback_to_user, reply_content)
             async with message_lock:
                 await ai_post_text_split_async_async(request, token_id, app_id, callback_to_user, reply_content)
diff --git a/app/main.py b/app/main.py
index 3258f7d..946a19b 100644
--- a/app/main.py
+++ b/app/main.py
@@ -161,6 +161,8 @@ async def lifespan(app: FastAPI):
     # # 初始化 GeWeChatCom 
     #app.state.gwechat_service = GeWeService(app)
 
+    app.state.message_lock={}
+
     # 初始化业务服务
     biz_service = BizService(app)
     app.state.biz_service = biz_service