update protocol to support manual response mode

main/protocols/protocol.h

@@ -12,6 +12,16 @@ struct BinaryProtocol3 {
     uint8_t payload[];
 } __attribute__((packed));
 
+enum AbortReason {
+    kAbortReasonNone,
+    kAbortReasonWakeWordDetected
+};
+
+enum ListeningMode {
+    kListeningModeAutoStop,
+    kListeningModeManualStop,
+    kListeningModeAlwaysOn // 需要 AEC 支持
+};
 
 class Protocol {
 public:
@@ -27,13 +37,14 @@ public:
     void OnAudioChannelClosed(std::function<void()> callback);
     void OnNetworkError(std::function<void(const std::string& message)> callback);
 
-    virtual void SendAudio(const std::string& data) = 0;
-    virtual void SendText(const std::string& text) = 0;
-    virtual void SendState(const std::string& state) = 0;
-    virtual void SendAbort() = 0;
     virtual bool OpenAudioChannel() = 0;
     virtual void CloseAudioChannel() = 0;
     virtual bool IsAudioChannelOpened() const = 0;
+    virtual void SendAudio(const std::string& data) = 0;
+    virtual void SendWakeWordDetected(const std::string& wake_word);
+    virtual void SendStartListening(ListeningMode mode);
+    virtual void SendStopListening();
+    virtual void SendAbortSpeaking(AbortReason reason);
 
 protected:
     std::function<void(const cJSON* root)> on_incoming_json_;
@@ -43,6 +54,9 @@ protected:
     std::function<void(const std::string& message)> on_network_error_;
 
     int server_sample_rate_ = 16000;
+    std::string session_id_;
+
+    virtual void SendText(const std::string& text) = 0;
 };
 
 #endif // PROTOCOL_H