diff --git a/main/application.cc b/main/application.cc
index c223054f..a0cab784 100644
--- a/main/application.cc
+++ b/main/application.cc
@@ -490,7 +490,7 @@ void Application::Start() {
             }
         }
     });
-    protocol_->Start();
+    bool protocol_started = protocol_->Start();
 
 #if CONFIG_USE_AUDIO_PROCESSOR
     audio_processor_.Initialize(codec, realtime_chat_enabled_);
@@ -556,12 +556,15 @@ void Application::Start() {
     // Wait for the new version check to finish
     xEventGroupWaitBits(event_group_, CHECK_NEW_VERSION_DONE_EVENT, pdTRUE, pdFALSE, portMAX_DELAY);
     SetDeviceState(kDeviceStateIdle);
-    std::string message = std::string(Lang::Strings::VERSION) + ota_.GetCurrentVersion();
-    display->ShowNotification(message.c_str());
-    display->SetChatMessage("system", "");
-    // Play the success sound to indicate the device is ready
-    ResetDecoder();
-    PlaySound(Lang::Sounds::P3_SUCCESS);
+
+    if (protocol_started) {
+        std::string message = std::string(Lang::Strings::VERSION) + ota_.GetCurrentVersion();
+        display->ShowNotification(message.c_str());
+        display->SetChatMessage("system", "");
+        // Play the success sound to indicate the device is ready
+        ResetDecoder();
+        PlaySound(Lang::Sounds::P3_SUCCESS);
+    }
     
     // Enter the main event loop
     MainEventLoop();
diff --git a/main/protocols/mqtt_protocol.cc b/main/protocols/mqtt_protocol.cc
index d5ac5635..3cda2e79 100644
--- a/main/protocols/mqtt_protocol.cc
+++ b/main/protocols/mqtt_protocol.cc
@@ -27,8 +27,8 @@ MqttProtocol::~MqttProtocol() {
     vEventGroupDelete(event_group_handle_);
 }
 
-void MqttProtocol::Start() {
-    StartMqttClient(false);
+bool MqttProtocol::Start() {
+    return StartMqttClient(false);
 }
 
 bool MqttProtocol::StartMqttClient(bool report_error) {
@@ -90,7 +90,16 @@ bool MqttProtocol::StartMqttClient(bool report_error) {
     });
 
     ESP_LOGI(TAG, "Connecting to endpoint %s", endpoint_.c_str());
-    if (!mqtt_->Connect(endpoint_, 8883, client_id_, username_, password_)) {
+    std::string broker_address;
+    int broker_port = 8883;
+    size_t pos = endpoint_.find(':');
+    if (pos != std::string::npos) {
+        broker_address = endpoint_.substr(0, pos);
+        broker_port = std::stoi(endpoint_.substr(pos + 1));
+    } else {
+        broker_address = endpoint_;
+    }
+    if (!mqtt_->Connect(broker_address, broker_port, client_id_, username_, password_)) {
         ESP_LOGE(TAG, "Failed to connect to endpoint");
         SetError(Lang::Strings::SERVER_NOT_CONNECTED);
         return false;
diff --git a/main/protocols/mqtt_protocol.h b/main/protocols/mqtt_protocol.h
index 8c75cd7b..e531b00a 100644
--- a/main/protocols/mqtt_protocol.h
+++ b/main/protocols/mqtt_protocol.h
@@ -25,7 +25,7 @@ public:
     MqttProtocol();
     ~MqttProtocol();
 
-    void Start() override;
+    bool Start() override;
     void SendAudio(const std::vector<uint8_t>& data) override;
     bool OpenAudioChannel() override;
     void CloseAudioChannel() override;
diff --git a/main/protocols/protocol.h b/main/protocols/protocol.h
index 1dbd268c..5fdd5751 100644
--- a/main/protocols/protocol.h
+++ b/main/protocols/protocol.h
@@ -44,7 +44,7 @@ public:
     void OnAudioChannelClosed(std::function<void()> callback);
     void OnNetworkError(std::function<void(const std::string& message)> callback);
 
-    virtual void Start() = 0;
+    virtual bool Start() = 0;
     virtual bool OpenAudioChannel() = 0;
     virtual void CloseAudioChannel() = 0;
     virtual bool IsAudioChannelOpened() const = 0;
diff --git a/main/protocols/websocket_protocol.cc b/main/protocols/websocket_protocol.cc
index 64e0740a..50b234a4 100644
--- a/main/protocols/websocket_protocol.cc
+++ b/main/protocols/websocket_protocol.cc
@@ -22,7 +22,9 @@ WebsocketProtocol::~WebsocketProtocol() {
     vEventGroupDelete(event_group_handle_);
 }
 
-void WebsocketProtocol::Start() {
+bool WebsocketProtocol::Start() {
+    // Only connect to server when audio channel is needed
+    return true;
 }
 
 void WebsocketProtocol::SendAudio(const std::vector<uint8_t>& data) {
diff --git a/main/protocols/websocket_protocol.h b/main/protocols/websocket_protocol.h
index 3f32e726..5e96083f 100644
--- a/main/protocols/websocket_protocol.h
+++ b/main/protocols/websocket_protocol.h
@@ -15,7 +15,7 @@ public:
     WebsocketProtocol();
     ~WebsocketProtocol();
 
-    void Start() override;
+    bool Start() override;
     void SendAudio(const std::vector<uint8_t>& data) override;
     bool OpenAudioChannel() override;
     void CloseAudioChannel() override;