diff --git a/.vscode/settings.json b/.vscode/settings.json
index c02be26..b7cce9f 100644
--- a/.vscode/settings.json
+++ b/.vscode/settings.json
@@ -11,6 +11,7 @@
         "istream": "cpp",
         "numeric": "cpp",
         "ostream": "cpp",
-        "sstream": "cpp"
+        "sstream": "cpp",
+        "chrono": "cpp"
     }
 }
diff --git a/README.md b/README.md
index 1d5e197..6ece298 100644
--- a/README.md
+++ b/README.md
@@ -88,6 +88,14 @@ async def talk_session(proxy: WsProxy):
             await proxy.speak(resp.text)
 ```
 
+`StackChanApp()` は既定で、WebSocket 接続直後に WakeWord 検出通知音をデバイスへ送信しようとします。
+送信する音は環境変数 `STACKCHAN_WAKEWORD_SOUND_PATH` で指定した WAV ファイルから読み込みます。
+読み込んだ WAV は送信前に 16-bit PCM / 24kHz / mono へ正規化されます。
+さらに短い通知音でも再生しやすいよう、送信前に前後へ短い無音を付与し、最小再生長を確保します。
+この値が未設定なら通知音は送信されません。
+送信された音はデバイス側で SPIFFS に保存され、WakeWord 検出時にローカル再生されます。
+接続時送信の機能自体を無効化したい場合は `StackChanApp(send_wakeword_sound_on_connect=False)` を使ってください。
+
 ## セットアップ
 
 以下を確認ください。
@@ -101,7 +109,7 @@ async def talk_session(proxy: WsProxy):
     - M5Stack CoreS3(SKU:K128, K128-Lite, K128-SE)
     - M5Stack Atom S3R(SKU:C126) + Atomic Echo Base(SKU:A149)
     - M5Stack公式StackChan(SKU:K151)
-    <!-- - M5Stack Atom EchoS3R -->
+    - M5Stack Atom EchoS3R
 - サーボ（なくても動作します）:
     - Tower Pro SG90
     - FEETECH SCS0009
diff --git a/docs/websocket_protocols_ja.md b/docs/websocket_protocols_ja.md
index b816c13..ee765e2 100644
--- a/docs/websocket_protocols_ja.md
+++ b/docs/websocket_protocols_ja.md
@@ -35,6 +35,7 @@
 | `SpeakDoneEvt` | CoreS3 → Server | 音声再生完了通知 |
 | `ServoCmd` | Server → CoreS3 | サーボ動作シーケンス指示 |
 | `ServoDoneEvt` | CoreS3 → Server | サーボ動作完了通知 |
+| `StoredFile` | Server → CoreS3 | SPIFFS 保存用の汎用ファイル転送 |
 
 ### `MessageType` 一覧
 
@@ -136,6 +137,29 @@
 - CoreS3 側の音声再生完了を通知します。
 - Server はこの通知を待って `proxy.speak()` を完了させます。
 
+## 保存ファイル転送 `StoredFile`
+
+- 方向: Server → CoreS3
+- 用途: バイナリファイルを WebSocket 経由で配布し、CoreS3 側で SPIFFS に保存するための汎用転送です。
+- 1 転送の流れは `StoredFileStart` → `FileChunk` 複数回 → `StoredFileEnd` です。
+
+### body 形式
+
+| messageType | body |
+| --- | --- |
+| `START` | `StoredFileStart { file_id, content_type, total_size, sample_rate, channels }` |
+| `DATA` | `FileChunk { chunk_bytes }` |
+| `END` | `StoredFileEnd {}` |
+
+### 現行実装メモ
+
+- `file_id` はファイルの論理名です。現在の実装では `wakeword-detected-sound` が WakeUpWord 検出音に使われます。
+- `content_type` は現在 `audio/pcm` をサポートします。
+- `sample_rate` / `channels` は PCM 再生用の追加メタデータです。
+- CoreS3 は受信完了後に SPIFFS へ保存し、**その WebSocket 接続中に受信したファイルだけ** を有効化します。
+- 再接続後にサーバーが同じファイルを再送しない場合、SPIFFS に過去データが残っていても再生には使いません。
+- WakeUpWord 検出時は、該当サウンドが現在の接続で有効になっている場合のみローカル再生してから `WakeWordEvt` を送信します。
+
 ## サーボ動作指示 `ServoCmd`
 
 - 方向: Server → CoreS3
diff --git a/firmware/include/protocols.hpp b/firmware/include/protocols.hpp
index 89cb7d6..c5de57f 100644
--- a/firmware/include/protocols.hpp
+++ b/firmware/include/protocols.hpp
@@ -8,6 +8,7 @@
 #include "../lib/generated_protobuf/websocket-message.pb.h"
 
 constexpr size_t kProtoAudioChunkMaxBytes = 4096;
+constexpr size_t kProtoFileChunkMaxBytes = 4096;
 constexpr size_t kProtoServoCommandMaxCount = 255;
 constexpr size_t kMaxEncodedWebSocketMessageBytes = stackchan_websocket_v1_WebSocketMessage_size;
 
@@ -18,6 +19,13 @@ bool setProtoAudioChunk(
 const uint8_t *getProtoAudioChunkBytes(const stackchan_websocket_v1_AudioChunk &chunk);
 size_t getProtoAudioChunkSize(const stackchan_websocket_v1_AudioChunk &chunk);
 
+bool setProtoFileChunk(
+	stackchan_websocket_v1_FileChunk &chunk,
+	const uint8_t *data,
+	size_t data_len);
+const uint8_t *getProtoFileChunkBytes(const stackchan_websocket_v1_FileChunk &chunk);
+size_t getProtoFileChunkSize(const stackchan_websocket_v1_FileChunk &chunk);
+
 bool encodeWebSocketMessage(
 	const stackchan_websocket_v1_WebSocketMessage &message,
 	std::vector<uint8_t> &encoded);
diff --git a/firmware/include/stored_files.hpp b/firmware/include/stored_files.hpp
new file mode 100644
index 0000000..7778d56
--- /dev/null
+++ b/firmware/include/stored_files.hpp
@@ -0,0 +1,73 @@
+#pragma once
+
+#include <array>
+#include <cstddef>
+#include <cstdint>
+#include <vector>
+
+#include "protocols.hpp"
+
+struct StoredFileView
+{
+  const uint8_t *data = nullptr;
+  size_t size = 0;
+  uint32_t sample_rate = 0;
+  uint16_t channels = 0;
+};
+
+class StoredFiles
+{
+public:
+  void init();
+  void resetSession();
+
+  bool handleStart(uint32_t seq, const stackchan_websocket_v1_StoredFileStart &start);
+  bool handleData(uint32_t seq, const uint8_t *data, size_t data_len);
+  bool handleEnd(uint32_t seq);
+
+  bool getActivePcmFile(const char *fileId, StoredFileView &view);
+
+private:
+  static constexpr size_t kMaxStoredFiles = 4;
+  static constexpr size_t kMaxStoredFileBytes = 256 * 1024;
+
+  struct PersistedSlot
+  {
+    bool used = false;
+    char file_id[64] = "";
+    char content_type[64] = "";
+    uint32_t sample_rate = 0;
+    uint32_t channels = 0;
+    uint32_t size = 0;
+  };
+
+  struct TransferState
+  {
+    bool active = false;
+    uint32_t next_seq = 0;
+    int slot_index = -1;
+    uint32_t received_bytes = 0;
+    uint32_t chunk_count = 0;
+    PersistedSlot slot{};
+    std::vector<uint8_t> payload;
+  };
+
+  bool storage_ready_ = false;
+  std::array<PersistedSlot, kMaxStoredFiles> slots_{};
+  std::array<bool, kMaxStoredFiles> session_active_{};
+  int cached_slot_index_ = -1;
+  std::vector<uint8_t> cached_payload_;
+  TransferState transfer_;
+
+  bool mountSpiffs();
+  bool loadIndex();
+  bool persistIndex();
+  bool persistSlotPayload(int slotIndex, const std::vector<uint8_t> &payload);
+  bool loadSlotPayload(int slotIndex, std::vector<uint8_t> &payload);
+  int findSlotById(const char *fileId) const;
+  int selectSlotForId(const char *fileId);
+  void resetTransfer();
+  bool activateSlot(int slotIndex, const std::vector<uint8_t> &payload);
+  static const char *payloadPathForSlot(int slotIndex);
+  static const char *indexPath();
+};
diff --git a/firmware/lib/generated_protobuf/websocket-message.pb.c b/firmware/lib/generated_protobuf/websocket-message.pb.c
index f70a79e..2d976af 100644
--- a/firmware/lib/generated_protobuf/websocket-message.pb.c
+++ b/firmware/lib/generated_protobuf/websocket-message.pb.c
@@ -24,6 +24,9 @@ PB_BIND(stackchan_websocket_v1_AudioWavEnd, stackchan_websocket_v1_AudioWavEnd,
 PB_BIND(stackchan_websocket_v1_AudioChunk, stackchan_websocket_v1_AudioChunk, 4)
 
 
+PB_BIND(stackchan_websocket_v1_FileChunk, stackchan_websocket_v1_FileChunk, 4)
+
+
 PB_BIND(stackchan_websocket_v1_StateCommand, stackchan_websocket_v1_StateCommand, AUTO)
 
 
@@ -36,6 +39,12 @@ PB_BIND(stackchan_websocket_v1_StateEvent, stackchan_websocket_v1_StateEvent, AU
 PB_BIND(stackchan_websocket_v1_SpeakDoneEvent, stackchan_websocket_v1_SpeakDoneEvent, AUTO)
 
 
+PB_BIND(stackchan_websocket_v1_StoredFileStart, stackchan_websocket_v1_StoredFileStart, AUTO)
+
+
+PB_BIND(stackchan_websocket_v1_StoredFileEnd, stackchan_websocket_v1_StoredFileEnd, AUTO)
+
+
 PB_BIND(stackchan_websocket_v1_ServoCommandSequence, stackchan_websocket_v1_ServoCommandSequence, 2)
 
 
diff --git a/firmware/lib/generated_protobuf/websocket-message.pb.h b/firmware/lib/generated_protobuf/websocket-message.pb.h
index 8e0c222..a668a7e 100644
--- a/firmware/lib/generated_protobuf/websocket-message.pb.h
+++ b/firmware/lib/generated_protobuf/websocket-message.pb.h
@@ -21,7 +21,8 @@ typedef enum _stackchan_websocket_v1_MessageKind {
     stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVO_CMD = 7,
     stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVO_DONE_EVT = 8,
     stackchan_websocket_v1_MessageKind_MESSAGE_KIND_FIRMWARE_METADATA = 9,
-    stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVER_METADATA = 10
+    stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVER_METADATA = 10,
+    stackchan_websocket_v1_MessageKind_MESSAGE_KIND_STORED_FILE = 11
 } stackchan_websocket_v1_MessageKind;
 
 typedef enum _stackchan_websocket_v1_MessageType {
@@ -81,6 +82,11 @@ typedef struct _stackchan_websocket_v1_AudioChunk {
     stackchan_websocket_v1_AudioChunk_pcm_bytes_t pcm_bytes;
 } stackchan_websocket_v1_AudioChunk;
 
+typedef PB_BYTES_ARRAY_T(4096) stackchan_websocket_v1_FileChunk_chunk_bytes_t;
+typedef struct _stackchan_websocket_v1_FileChunk {
+    stackchan_websocket_v1_FileChunk_chunk_bytes_t chunk_bytes;
+} stackchan_websocket_v1_FileChunk;
+
 typedef struct _stackchan_websocket_v1_StateCommand {
     stackchan_websocket_v1_StackchanState state;
 } stackchan_websocket_v1_StateCommand;
@@ -97,6 +103,18 @@ typedef struct _stackchan_websocket_v1_SpeakDoneEvent {
     bool done;
 } stackchan_websocket_v1_SpeakDoneEvent;
 
+typedef struct _stackchan_websocket_v1_StoredFileStart {
+    char file_id[64];
+    char content_type[64];
+    uint32_t total_size;
+    uint32_t sample_rate;
+    uint32_t channels;
+} stackchan_websocket_v1_StoredFileStart;
+
+typedef struct _stackchan_websocket_v1_StoredFileEnd {
+    char dummy_field;
+} stackchan_websocket_v1_StoredFileEnd;
+
 typedef struct _stackchan_websocket_v1_ServoCommand {
     stackchan_websocket_v1_ServoOperation op;
     int32_t angle; /* used by MOVE_X / MOVE_Y */
@@ -155,6 +173,9 @@ typedef struct _stackchan_websocket_v1_WebSocketMessage {
         stackchan_websocket_v1_ServoDoneEvent servo_done_evt;
         stackchan_websocket_v1_FirmwareMetadata firmware_metadata;
         stackchan_websocket_v1_ServerMetadata server_metadata;
+        stackchan_websocket_v1_StoredFileStart stored_file_start;
+        stackchan_websocket_v1_FileChunk stored_file_data;
+        stackchan_websocket_v1_StoredFileEnd stored_file_end;
     } body;
 } stackchan_websocket_v1_WebSocketMessage;
 
@@ -165,8 +186,8 @@ extern "C" {
 
 /* Helper constants for enums */
 #define _stackchan_websocket_v1_MessageKind_MIN stackchan_websocket_v1_MessageKind_MESSAGE_KIND_UNSPECIFIED
-#define _stackchan_websocket_v1_MessageKind_MAX stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVER_METADATA
-#define _stackchan_websocket_v1_MessageKind_ARRAYSIZE ((stackchan_websocket_v1_MessageKind)(stackchan_websocket_v1_MessageKind_MESSAGE_KIND_SERVER_METADATA+1))
+#define _stackchan_websocket_v1_MessageKind_MAX stackchan_websocket_v1_MessageKind_MESSAGE_KIND_STORED_FILE
+#define _stackchan_websocket_v1_MessageKind_ARRAYSIZE ((stackchan_websocket_v1_MessageKind)(stackchan_websocket_v1_MessageKind_MESSAGE_KIND_STORED_FILE+1))
 
 #define _stackchan_websocket_v1_MessageType_MIN stackchan_websocket_v1_MessageType_MESSAGE_TYPE_UNSPECIFIED
 #define _stackchan_websocket_v1_MessageType_MAX stackchan_websocket_v1_MessageType_MESSAGE_TYPE_END
@@ -196,6 +217,7 @@ extern "C" {
 
 
 
+
 #define stackchan_websocket_v1_StateCommand_state_ENUMTYPE stackchan_websocket_v1_StackchanState
 
 
@@ -203,6 +225,8 @@ extern "C" {
 
 
 
+
+
 #define stackchan_websocket_v1_ServoCommand_op_ENUMTYPE stackchan_websocket_v1_ServoOperation
 
 
@@ -218,10 +242,13 @@ extern "C" {
 #define stackchan_websocket_v1_AudioWavStart_init_default {0, 0}
 #define stackchan_websocket_v1_AudioWavEnd_init_default {0}
 #define stackchan_websocket_v1_AudioChunk_init_default {{0, {0}}}
+#define stackchan_websocket_v1_FileChunk_init_default {{0, {0}}}
 #define stackchan_websocket_v1_StateCommand_init_default {_stackchan_websocket_v1_StackchanState_MIN}
 #define stackchan_websocket_v1_WakeWordEvent_init_default {0}
 #define stackchan_websocket_v1_StateEvent_init_default {_stackchan_websocket_v1_StackchanState_MIN}
 #define stackchan_websocket_v1_SpeakDoneEvent_init_default {0}
+#define stackchan_websocket_v1_StoredFileStart_init_default {"", "", 0, 0, 0}
+#define stackchan_websocket_v1_StoredFileEnd_init_default {0}
 #define stackchan_websocket_v1_ServoCommandSequence_init_default {0, {stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default, stackchan_websocket_v1_ServoCommand_init_default}}
 #define stackchan_websocket_v1_ServoCommand_init_default {_stackchan_websocket_v1_ServoOperation_MIN, 0, 0}
 #define stackchan_websocket_v1_ServoDoneEvent_init_default {0}
@@ -233,10 +260,13 @@ extern "C" {
 #define stackchan_websocket_v1_AudioWavStart_init_zero {0, 0}
 #define stackchan_websocket_v1_AudioWavEnd_init_zero {0}
 #define stackchan_websocket_v1_AudioChunk_init_zero {{0, {0}}}
+#define stackchan_websocket_v1_FileChunk_init_zero {{0, {0}}}
 #define stackchan_websocket_v1_StateCommand_init_zero {_stackchan_websocket_v1_StackchanState_MIN}
 #define stackchan_websocket_v1_WakeWordEvent_init_zero {0}
 #define stackchan_websocket_v1_StateEvent_init_zero {_stackchan_websocket_v1_StackchanState_MIN}
 #define stackchan_websocket_v1_SpeakDoneEvent_init_zero {0}
+#define stackchan_websocket_v1_StoredFileStart_init_zero {"", "", 0, 0, 0}
+#define stackchan_websocket_v1_StoredFileEnd_init_zero {0}
 #define stackchan_websocket_v1_ServoCommandSequence_init_zero {0, {stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero, stackchan_websocket_v1_ServoCommand_init_zero}}
 #define stackchan_websocket_v1_ServoCommand_init_zero {_stackchan_websocket_v1_ServoOperation_MIN, 0, 0}
 #define stackchan_websocket_v1_ServoDoneEvent_init_zero {0}
@@ -247,10 +277,16 @@ extern "C" {
 #define stackchan_websocket_v1_AudioWavStart_sample_rate_tag 1
 #define stackchan_websocket_v1_AudioWavStart_channels_tag 2
 #define stackchan_websocket_v1_AudioChunk_pcm_bytes_tag 1
+#define stackchan_websocket_v1_FileChunk_chunk_bytes_tag 1
 #define stackchan_websocket_v1_StateCommand_state_tag 1
 #define stackchan_websocket_v1_WakeWordEvent_detected_tag 1
 #define stackchan_websocket_v1_StateEvent_state_tag 1
 #define stackchan_websocket_v1_SpeakDoneEvent_done_tag 1
+#define stackchan_websocket_v1_StoredFileStart_file_id_tag 1
+#define stackchan_websocket_v1_StoredFileStart_content_type_tag 2
+#define stackchan_websocket_v1_StoredFileStart_total_size_tag 3
+#define stackchan_websocket_v1_StoredFileStart_sample_rate_tag 4
+#define stackchan_websocket_v1_StoredFileStart_channels_tag 5
 #define stackchan_websocket_v1_ServoCommand_op_tag 1
 #define stackchan_websocket_v1_ServoCommand_angle_tag 2
 #define stackchan_websocket_v1_ServoCommand_duration_ms_tag 3
@@ -283,6 +319,9 @@ extern "C" {
 #define stackchan_websocket_v1_WebSocketMessage_servo_done_evt_tag 35
 #define stackchan_websocket_v1_WebSocketMessage_firmware_metadata_tag 36
 #define stackchan_websocket_v1_WebSocketMessage_server_metadata_tag 37
+#define stackchan_websocket_v1_WebSocketMessage_stored_file_start_tag 40
+#define stackchan_websocket_v1_WebSocketMessage_stored_file_data_tag 41
+#define stackchan_websocket_v1_WebSocketMessage_stored_file_end_tag 42
 
 /* Struct field encoding specification for nanopb */
 #define stackchan_websocket_v1_WebSocketMessage_FIELDLIST(X, a) \
@@ -302,7 +341,10 @@ X(a, STATIC,   ONEOF,    MESSAGE,  (body,speak_done_evt,body.speak_done_evt),  3
 X(a, STATIC,   ONEOF,    MESSAGE,  (body,servo_cmd,body.servo_cmd),  34) \
 X(a, STATIC,   ONEOF,    MESSAGE,  (body,servo_done_evt,body.servo_done_evt),  35) \
 X(a, STATIC,   ONEOF,    MESSAGE,  (body,firmware_metadata,body.firmware_metadata),  36) \
-X(a, STATIC,   ONEOF,    MESSAGE,  (body,server_metadata,body.server_metadata),  37)
+X(a, STATIC,   ONEOF,    MESSAGE,  (body,server_metadata,body.server_metadata),  37) \
+X(a, STATIC,   ONEOF,    MESSAGE,  (body,stored_file_start,body.stored_file_start),  40) \
+X(a, STATIC,   ONEOF,    MESSAGE,  (body,stored_file_data,body.stored_file_data),  41) \
+X(a, STATIC,   ONEOF,    MESSAGE,  (body,stored_file_end,body.stored_file_end),  42)
 #define stackchan_websocket_v1_WebSocketMessage_CALLBACK NULL
 #define stackchan_websocket_v1_WebSocketMessage_DEFAULT NULL
 #define stackchan_websocket_v1_WebSocketMessage_body_audio_pcm_start_MSGTYPE stackchan_websocket_v1_AudioPcmStart
@@ -319,6 +361,9 @@ X(a, STATIC,   ONEOF,    MESSAGE,  (body,server_metadata,body.server_metadata),
 #define stackchan_websocket_v1_WebSocketMessage_body_servo_done_evt_MSGTYPE stackchan_websocket_v1_ServoDoneEvent
 #define stackchan_websocket_v1_WebSocketMessage_body_firmware_metadata_MSGTYPE stackchan_websocket_v1_FirmwareMetadata
 #define stackchan_websocket_v1_WebSocketMessage_body_server_metadata_MSGTYPE stackchan_websocket_v1_ServerMetadata
+#define stackchan_websocket_v1_WebSocketMessage_body_stored_file_start_MSGTYPE stackchan_websocket_v1_StoredFileStart
+#define stackchan_websocket_v1_WebSocketMessage_body_stored_file_data_MSGTYPE stackchan_websocket_v1_FileChunk
+#define stackchan_websocket_v1_WebSocketMessage_body_stored_file_end_MSGTYPE stackchan_websocket_v1_StoredFileEnd
 
 #define stackchan_websocket_v1_AudioPcmStart_FIELDLIST(X, a) \
 
@@ -346,6 +391,11 @@ X(a, STATIC,   SINGULAR, BYTES,    pcm_bytes,         1)
 #define stackchan_websocket_v1_AudioChunk_CALLBACK NULL
 #define stackchan_websocket_v1_AudioChunk_DEFAULT NULL
 
+#define stackchan_websocket_v1_FileChunk_FIELDLIST(X, a) \
+X(a, STATIC,   SINGULAR, BYTES,    chunk_bytes,       1)
+#define stackchan_websocket_v1_FileChunk_CALLBACK NULL
+#define stackchan_websocket_v1_FileChunk_DEFAULT NULL
+
 #define stackchan_websocket_v1_StateCommand_FIELDLIST(X, a) \
 X(a, STATIC,   SINGULAR, UENUM,    state,             1)
 #define stackchan_websocket_v1_StateCommand_CALLBACK NULL
@@ -366,6 +416,20 @@ X(a, STATIC,   SINGULAR, BOOL,     done,              1)
 #define stackchan_websocket_v1_SpeakDoneEvent_CALLBACK NULL
 #define stackchan_websocket_v1_SpeakDoneEvent_DEFAULT NULL
 
+#define stackchan_websocket_v1_StoredFileStart_FIELDLIST(X, a) \
+X(a, STATIC,   SINGULAR, STRING,   file_id,           1) \
+X(a, STATIC,   SINGULAR, STRING,   content_type,      2) \
+X(a, STATIC,   SINGULAR, UINT32,   total_size,        3) \
+X(a, STATIC,   SINGULAR, UINT32,   sample_rate,       4) \
+X(a, STATIC,   SINGULAR, UINT32,   channels,          5)
+#define stackchan_websocket_v1_StoredFileStart_CALLBACK NULL
+#define stackchan_websocket_v1_StoredFileStart_DEFAULT NULL
+
+#define stackchan_websocket_v1_StoredFileEnd_FIELDLIST(X, a) \
+
+#define stackchan_websocket_v1_StoredFileEnd_CALLBACK NULL
+#define stackchan_websocket_v1_StoredFileEnd_DEFAULT NULL
+
 #define stackchan_websocket_v1_ServoCommandSequence_FIELDLIST(X, a) \
 X(a, STATIC,   REPEATED, MESSAGE,  commands,          1)
 #define stackchan_websocket_v1_ServoCommandSequence_CALLBACK NULL
@@ -408,10 +472,13 @@ extern const pb_msgdesc_t stackchan_websocket_v1_AudioPcmEnd_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_AudioWavStart_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_AudioWavEnd_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_AudioChunk_msg;
+extern const pb_msgdesc_t stackchan_websocket_v1_FileChunk_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_StateCommand_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_WakeWordEvent_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_StateEvent_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_SpeakDoneEvent_msg;
+extern const pb_msgdesc_t stackchan_websocket_v1_StoredFileStart_msg;
+extern const pb_msgdesc_t stackchan_websocket_v1_StoredFileEnd_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_ServoCommandSequence_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_ServoCommand_msg;
 extern const pb_msgdesc_t stackchan_websocket_v1_ServoDoneEvent_msg;
@@ -425,10 +492,13 @@ extern const pb_msgdesc_t stackchan_websocket_v1_ServerMetadata_msg;
 #define stackchan_websocket_v1_AudioWavStart_fields &stackchan_websocket_v1_AudioWavStart_msg
 #define stackchan_websocket_v1_AudioWavEnd_fields &stackchan_websocket_v1_AudioWavEnd_msg
 #define stackchan_websocket_v1_AudioChunk_fields &stackchan_websocket_v1_AudioChunk_msg
+#define stackchan_websocket_v1_FileChunk_fields &stackchan_websocket_v1_FileChunk_msg
 #define stackchan_websocket_v1_StateCommand_fields &stackchan_websocket_v1_StateCommand_msg
 #define stackchan_websocket_v1_WakeWordEvent_fields &stackchan_websocket_v1_WakeWordEvent_msg
 #define stackchan_websocket_v1_StateEvent_fields &stackchan_websocket_v1_StateEvent_msg
 #define stackchan_websocket_v1_SpeakDoneEvent_fields &stackchan_websocket_v1_SpeakDoneEvent_msg
+#define stackchan_websocket_v1_StoredFileStart_fields &stackchan_websocket_v1_StoredFileStart_msg
+#define stackchan_websocket_v1_StoredFileEnd_fields &stackchan_websocket_v1_StoredFileEnd_msg
 #define stackchan_websocket_v1_ServoCommandSequence_fields &stackchan_websocket_v1_ServoCommandSequence_msg
 #define stackchan_websocket_v1_ServoCommand_fields &stackchan_websocket_v1_ServoCommand_msg
 #define stackchan_websocket_v1_ServoDoneEvent_fields &stackchan_websocket_v1_ServoDoneEvent_msg
@@ -442,6 +512,7 @@ extern const pb_msgdesc_t stackchan_websocket_v1_ServerMetadata_msg;
 #define stackchan_websocket_v1_AudioPcmStart_size 0
 #define stackchan_websocket_v1_AudioWavEnd_size  0
 #define stackchan_websocket_v1_AudioWavStart_size 12
+#define stackchan_websocket_v1_FileChunk_size    4099
 #define stackchan_websocket_v1_FirmwareMetadata_size 87
 #define stackchan_websocket_v1_ServerMetadata_size 67
 #define stackchan_websocket_v1_ServoCommandSequence_size 4080
@@ -450,6 +521,8 @@ extern const pb_msgdesc_t stackchan_websocket_v1_ServerMetadata_msg;
 #define stackchan_websocket_v1_SpeakDoneEvent_size 2
 #define stackchan_websocket_v1_StateCommand_size 2
 #define stackchan_websocket_v1_StateEvent_size   2
+#define stackchan_websocket_v1_StoredFileEnd_size 0
+#define stackchan_websocket_v1_StoredFileStart_size 148
 #define stackchan_websocket_v1_WakeWordEvent_size 2
 #define stackchan_websocket_v1_WebSocketMessage_size 4113
 
diff --git a/firmware/src/listening.cpp b/firmware/src/listening.cpp
index edb2e35..d7e6aa6 100644
--- a/firmware/src/listening.cpp
+++ b/firmware/src/listening.cpp
@@ -41,6 +41,12 @@ void Listening::init()
 
 void Listening::begin()
 {
+  if (M5.Speaker.isPlaying())
+  {
+    log_i("Stopping speaker playback before listening start");
+    M5.Speaker.stop();
+    delay(20);
+  }
   M5.Mic.begin();
   startStreaming();
 }
diff --git a/firmware/src/main.cpp b/firmware/src/main.cpp
index 6fb95c4..0d838d5 100644
--- a/firmware/src/main.cpp
+++ b/firmware/src/main.cpp
@@ -24,6 +24,7 @@
 #include "../include/wake_up_word.hpp"
 #include "../include/display.hpp"
 #include "../include/servo.hpp"
+#include "../include/stored_files.hpp"
 
 //////////////////// 設定 ////////////////////
 const char *WIFI_SSID = WIFI_SSID_H;
@@ -32,6 +33,11 @@ const char *SERVER_HOST = SERVER_HOST_H;
 const int SERVER_PORT = SERVER_PORT_H;
 const char *SERVER_PATH = SERVER_PATH_H; // WebSocket エンドポイント
 const int SAMPLE_RATE = 16000;           // 16kHz モノラル
+const int SPEAKER_OUTPUT_SAMPLE_RATE = 24000;
+const int SPEAKER_VOLUME = 160;          // 0-255
+const size_t SPEAKER_DMA_BUF_LEN = 512;
+const size_t SPEAKER_DMA_BUF_COUNT = 12;
+const uint8_t SPEAKER_TASK_PRIORITY = 3;
 /////////////////////////////////////////////
 
 StateMachine stateMachine;
@@ -42,6 +48,7 @@ static Listening listening(wsClient, stateMachine, SAMPLE_RATE);
 static WakeUpWord wakeUpWord(stateMachine, SAMPLE_RATE);
 static Display display(stateMachine);
 static BodyServo servo;
+static StoredFiles storedFiles;
 
 // Protocol types are defined in include/protocols.hpp
 namespace
@@ -51,8 +58,12 @@ uint32_t g_last_comm_ms = 0;
 uint32_t g_last_local_wake_word_ms = 0;
 constexpr uint32_t kCommTimeoutMs = 60000;
 constexpr uint32_t kLocalWakeWordCooldownMs = 750;
+constexpr const char *kWakeWordDetectedSoundFileId = "wakeword-detected-sound";
 stackchan_websocket_v1_WebSocketMessage g_tx_message = stackchan_websocket_v1_WebSocketMessage_init_zero;
 stackchan_websocket_v1_WebSocketMessage g_rx_message = stackchan_websocket_v1_WebSocketMessage_init_zero;
+bool g_pending_device_wake_word_feedback = false;
+bool g_feedback_sound_playing = false;
+bool g_notify_after_feedback_sound = false;
 
 void markCommunicationActive()
 {
@@ -156,6 +167,106 @@ void triggerLocalWakeWord(const char *source)
   notifyWakeWordDetected();
 }
 
+void clearFeedbackSoundState(bool stopPlayback)
+{
+  g_pending_device_wake_word_feedback = false;
+  g_notify_after_feedback_sound = false;
+  if (stopPlayback && g_feedback_sound_playing && M5.Speaker.isPlaying())
+  {
+    M5.Speaker.stop();
+  }
+  g_feedback_sound_playing = false;
+}
+
+bool startWakeWordFeedbackSoundPlayback()
+{
+  StoredFileView sound;
+  if (!storedFiles.getActivePcmFile(kWakeWordDetectedSoundFileId, sound))
+  {
+    log_w("Wake-word feedback sound is not available for current session id=%s", kWakeWordDetectedSoundFileId);
+    return false;
+  }
+
+  if (sound.size == 0 || (sound.size % sizeof(int16_t)) != 0)
+  {
+    log_w("Stored wake-word sound has invalid size=%u", static_cast<unsigned>(sound.size));
+    return false;
+  }
+
+  if (sound.sample_rate == 0 || sound.channels == 0)
+  {
+    log_w("Stored wake-word sound missing format sample_rate=%u channels=%u",
+          static_cast<unsigned>(sound.sample_rate),
+          static_cast<unsigned>(sound.channels));
+    return false;
+  }
+
+  wakeUpWord.end();
+  if (M5.Speaker.isPlaying())
+  {
+    M5.Speaker.stop();
+  }
+
+  const int16_t *samples = reinterpret_cast<const int16_t *>(sound.data);
+  size_t sample_len = sound.size / sizeof(int16_t);
+  log_i("Starting wake-word feedback playback sample_len=%u sample_rate=%u channels=%u",
+        static_cast<unsigned>(sample_len),
+        static_cast<unsigned>(sound.sample_rate),
+        static_cast<unsigned>(sound.channels));
+  M5.Speaker.playRaw(samples, sample_len, sound.sample_rate, sound.channels > 1, 1, 0);
+  g_feedback_sound_playing = true;
+  log_i("Playing stored wake-word sound bytes=%u", static_cast<unsigned>(sound.size));
+  return true;
+}
+
+void finishWakeWordFeedbackSoundPlayback()
+{
+  log_i("Wake-word feedback playback finished notify_after=%u",
+        static_cast<unsigned>(g_notify_after_feedback_sound));
+  g_feedback_sound_playing = false;
+  if (g_notify_after_feedback_sound)
+  {
+    g_notify_after_feedback_sound = false;
+    notifyWakeWordDetected();
+  }
+
+  if (stateMachine.getState() == StateMachine::Idle && shouldUseDeviceWakeWord())
+  {
+    wakeUpWord.begin();
+  }
+}
+
+void processPendingDeviceWakeWordFeedback()
+{
+  if (!g_pending_device_wake_word_feedback || stateMachine.getState() != StateMachine::Idle)
+  {
+    return;
+  }
+
+  g_pending_device_wake_word_feedback = false;
+  if (startWakeWordFeedbackSoundPlayback())
+  {
+    g_notify_after_feedback_sound = true;
+    log_i("Wake-word feedback playback scheduled before uplink event");
+    return;
+  }
+
+  log_w("Wake-word feedback playback skipped; sending WakeWordEvt immediately");
+  notifyWakeWordDetected();
+}
+
+void handleDeviceWakeWordDetected()
+{
+  if (!canTriggerLocalWakeWord())
+  {
+    return;
+  }
+
+  g_last_local_wake_word_ms = millis();
+  log_i("Local wake-word trigger from device wake word");
+  g_pending_device_wake_word_feedback = true;
+}
+
 void handleTouchWakeWordInput()
 {
 #if USE_STACKCHAN_BSP
@@ -324,11 +435,15 @@ void handleWsEvent(WStype_t type, uint8_t *payload, size_t length)
     // M5.Display.println("WS: disconnected");
     log_i("WS disconnected");
     resetServerMetadata();
+    storedFiles.resetSession();
+    clearFeedbackSoundState(true);
     stateMachine.setState(StateMachine::Disconnected);
     break;
   case WStype_CONNECTED:
     // M5.Display.printf("WS: connected %s\n", SERVER_PATH);
     log_i("WS connected to %s", SERVER_PATH);
+    storedFiles.resetSession();
+    clearFeedbackSoundState(true);
     if (stateMachine.getState() == StateMachine::Disconnected)
     {
       stateMachine.setState(StateMachine::Idle);
@@ -427,6 +542,49 @@ void handleWsEvent(WStype_t type, uint8_t *payload, size_t length)
         log_w("ServerMetadata protobuf body mismatch type=%u body=%u", (unsigned)rx.message_type, (unsigned)rx.which_body);
       }
       break;
+    case stackchan_websocket_v1_MessageKind_MESSAGE_KIND_STORED_FILE:
+      if (rx.message_type == stackchan_websocket_v1_MessageType_MESSAGE_TYPE_START &&
+          rx.which_body == stackchan_websocket_v1_WebSocketMessage_stored_file_start_tag)
+      {
+        log_i("Received stored file start id=%s seq=%u size=%u sample_rate=%u channels=%u",
+              rx.body.stored_file_start.file_id,
+              static_cast<unsigned>(rx.seq),
+              static_cast<unsigned>(rx.body.stored_file_start.total_size),
+              static_cast<unsigned>(rx.body.stored_file_start.sample_rate),
+              static_cast<unsigned>(rx.body.stored_file_start.channels));
+        if (!storedFiles.handleStart(rx.seq, rx.body.stored_file_start))
+        {
+          log_w("Stored file start rejected");
+        }
+      }
+      else if (rx.message_type == stackchan_websocket_v1_MessageType_MESSAGE_TYPE_DATA &&
+               rx.which_body == stackchan_websocket_v1_WebSocketMessage_stored_file_data_tag)
+      {
+        log_i("Received stored file chunk seq=%u bytes=%u",
+              static_cast<unsigned>(rx.seq),
+              static_cast<unsigned>(getProtoFileChunkSize(rx.body.stored_file_data)));
+        if (!storedFiles.handleData(
+                rx.seq,
+                getProtoFileChunkBytes(rx.body.stored_file_data),
+                getProtoFileChunkSize(rx.body.stored_file_data)))
+        {
+          log_w("Stored file data rejected");
+        }
+      }
+      else if (rx.message_type == stackchan_websocket_v1_MessageType_MESSAGE_TYPE_END &&
+               rx.which_body == stackchan_websocket_v1_WebSocketMessage_stored_file_end_tag)
+      {
+        log_i("Received stored file end seq=%u", static_cast<unsigned>(rx.seq));
+        if (!storedFiles.handleEnd(rx.seq))
+        {
+          log_w("Stored file end rejected");
+        }
+      }
+      else
+      {
+        log_w("StoredFile protobuf body mismatch type=%u body=%u", (unsigned)rx.message_type, (unsigned)rx.which_body);
+      }
+      break;
     default:
       // M5.Display.printf("WS bin kind=%u len=%d\n", (unsigned)rx.kind, (int)length);
       break;
@@ -459,6 +617,18 @@ void setup()
   M5.begin(cfg);
 #endif
 
+  auto spk_cfg = M5.Speaker.config();
+  spk_cfg.sample_rate = SPEAKER_OUTPUT_SAMPLE_RATE;
+  spk_cfg.dma_buf_len = SPEAKER_DMA_BUF_LEN;
+  spk_cfg.dma_buf_count = SPEAKER_DMA_BUF_COUNT;
+  spk_cfg.task_priority = SPEAKER_TASK_PRIORITY;
+  M5.Speaker.config(spk_cfg);
+  log_i("Speaker config sample_rate=%u dma_buf_len=%u dma_buf_count=%u task_priority=%u",
+        static_cast<unsigned>(spk_cfg.sample_rate),
+        static_cast<unsigned>(spk_cfg.dma_buf_len),
+        static_cast<unsigned>(spk_cfg.dma_buf_count),
+        static_cast<unsigned>(spk_cfg.task_priority));
+
   auto mic_cfg = M5.Mic.config();
   mic_cfg.sample_rate = SAMPLE_RATE;
   mic_cfg.dma_buf_len = 256;
@@ -468,6 +638,7 @@ void setup()
 
   listening.init();
   speaking.init();
+  storedFiles.init();
   speaking.setSpeakFinishedCallback([]() {
     notifySpeakDone();
   });
@@ -477,7 +648,7 @@ void setup()
   });
   wakeUpWord.init();
   wakeUpWord.setWakeWordDetectedCallback([]() {
-    notifyWakeWordDetected();
+    handleDeviceWakeWordDetected();
   });
   display.init();
   initializeFirmwareMetadata();
@@ -485,7 +656,7 @@ void setup()
   connectWiFi();
 
   // Mic/Speaking setup
-  M5.Speaker.setVolume(200); // 0-255
+  M5.Speaker.setVolume(SPEAKER_VOLUME); // 0-255
 
   wsClient.begin(SERVER_HOST, SERVER_PORT, SERVER_PATH);
   markCommunicationActive();
@@ -507,6 +678,7 @@ void setup()
 
   stateMachine.addStateEntryEvent(StateMachine::Listening, [](StateMachine::State, StateMachine::State) {
     notifyCurrentState(StateMachine::Listening);
+    clearFeedbackSoundState(true);
     listening.begin();
   });
   stateMachine.addStateExitEvent(StateMachine::Listening, [](StateMachine::State, StateMachine::State) {
@@ -541,11 +713,21 @@ void loop()
   switch (current)
   {
   case StateMachine::Idle:
+    if (g_feedback_sound_playing)
+    {
+      if (!M5.Speaker.isPlaying())
+      {
+        finishWakeWordFeedbackSoundPlayback();
+      }
+      break;
+    }
+
     handleTouchWakeWordInput();
     if (shouldUseDeviceWakeWord())
     {
       wakeUpWord.loop();
     }
+    processPendingDeviceWakeWordFeedback();
     break;
   case StateMachine::Listening:
     listening.loop();
diff --git a/firmware/src/protocols.cpp b/firmware/src/protocols.cpp
index 16af592..d6aa03b 100644
--- a/firmware/src/protocols.cpp
+++ b/firmware/src/protocols.cpp
@@ -33,6 +33,34 @@ size_t getProtoAudioChunkSize(const stackchan_websocket_v1_AudioChunk &chunk)
 	return chunk.pcm_bytes.size;
 }
 
+bool setProtoFileChunk(
+	stackchan_websocket_v1_FileChunk &chunk,
+	const uint8_t *data,
+	size_t data_len)
+{
+	if (data_len > kProtoFileChunkMaxBytes)
+	{
+		return false;
+	}
+
+	chunk.chunk_bytes.size = static_cast<pb_size_t>(data_len);
+	if (data_len > 0 && data != nullptr)
+	{
+		memcpy(chunk.chunk_bytes.bytes, data, data_len);
+	}
+	return true;
+}
+
+const uint8_t *getProtoFileChunkBytes(const stackchan_websocket_v1_FileChunk &chunk)
+{
+	return chunk.chunk_bytes.bytes;
+}
+
+size_t getProtoFileChunkSize(const stackchan_websocket_v1_FileChunk &chunk)
+{
+	return chunk.chunk_bytes.size;
+}
+
 bool encodeWebSocketMessage(
 	const stackchan_websocket_v1_WebSocketMessage &message,
 	std::vector<uint8_t> &encoded)
diff --git a/firmware/src/stored_files.cpp b/firmware/src/stored_files.cpp
new file mode 100644
index 0000000..dff459e
--- /dev/null
+++ b/firmware/src/stored_files.cpp
@@ -0,0 +1,494 @@
+#include "stored_files.hpp"
+
+#include <M5Unified.h>
+#include <SPIFFS.h>
+
+#include <cstdio>
+#include <cstring>
+#include <utility>
+
+namespace
+{
+constexpr const char *kIndexFilePath = "/wsfiles.idx";
+constexpr const char *kPayloadPaths[] = {
+    "/wsfile0.bin",
+    "/wsfile1.bin",
+    "/wsfile2.bin",
+    "/wsfile3.bin",
+};
+} // namespace
+
+void StoredFiles::init()
+{
+  session_active_.fill(false);
+  cached_slot_index_ = -1;
+  cached_payload_.clear();
+  resetTransfer();
+
+  if (!mountSpiffs())
+  {
+    return;
+  }
+
+  log_i(
+      "SPIFFS mounted total=%u used=%u",
+      static_cast<unsigned>(SPIFFS.totalBytes()),
+      static_cast<unsigned>(SPIFFS.usedBytes()));
+
+  if (!loadIndex())
+  {
+    slots_ = {};
+    persistIndex();
+  }
+}
+
+void StoredFiles::resetSession()
+{
+  session_active_.fill(false);
+  cached_slot_index_ = -1;
+  cached_payload_.clear();
+  resetTransfer();
+}
+
+bool StoredFiles::handleStart(uint32_t seq, const stackchan_websocket_v1_StoredFileStart &start)
+{
+  if (!mountSpiffs())
+  {
+    return false;
+  }
+
+  resetTransfer();
+
+  if (start.file_id[0] == '\0')
+  {
+    log_w("Stored file start missing file_id");
+    return false;
+  }
+
+  size_t total_size = static_cast<size_t>(start.total_size);
+  if (total_size > kMaxStoredFileBytes)
+  {
+    log_w(
+        "Stored file too large for SPIFFS-backed transfer id=%s size=%u limit=%u",
+        start.file_id,
+        static_cast<unsigned>(start.total_size),
+        static_cast<unsigned>(kMaxStoredFileBytes));
+    return false;
+  }
+
+  int slot_index = selectSlotForId(start.file_id);
+  if (slot_index < 0)
+  {
+    log_w("No slot available for stored file id=%s", start.file_id);
+    return false;
+  }
+
+  size_t reclaimable_bytes = slots_[slot_index].used ? slots_[slot_index].size : 0;
+  size_t total_bytes = SPIFFS.totalBytes();
+  size_t used_bytes = SPIFFS.usedBytes();
+  size_t free_bytes = total_bytes > used_bytes ? total_bytes - used_bytes : 0;
+  size_t available_bytes = free_bytes + reclaimable_bytes;
+  if (total_bytes > 0 && total_size > available_bytes)
+  {
+    log_w(
+        "Insufficient SPIFFS space for stored file id=%s requested=%u free=%u reclaimable=%u",
+        start.file_id,
+        static_cast<unsigned>(total_size),
+        static_cast<unsigned>(free_bytes),
+        static_cast<unsigned>(reclaimable_bytes));
+    return false;
+  }
+
+  transfer_.active = true;
+  transfer_.next_seq = seq + 1;
+  transfer_.slot_index = slot_index;
+  transfer_.slot = PersistedSlot{};
+  transfer_.slot.used = true;
+  snprintf(transfer_.slot.file_id, sizeof(transfer_.slot.file_id), "%s", start.file_id);
+  snprintf(transfer_.slot.content_type, sizeof(transfer_.slot.content_type), "%s", start.content_type);
+  transfer_.slot.sample_rate = start.sample_rate;
+  transfer_.slot.channels = start.channels;
+  transfer_.slot.size = static_cast<uint32_t>(total_size);
+  transfer_.received_bytes = 0;
+  transfer_.chunk_count = 0;
+  transfer_.payload.clear();
+  transfer_.payload.reserve(total_size);
+
+  log_i(
+      "Stored file start id=%s type=%s size=%u sample_rate=%u channels=%u slot=%d spiffs_free=%u",
+      transfer_.slot.file_id,
+      transfer_.slot.content_type,
+      static_cast<unsigned>(transfer_.slot.size),
+      static_cast<unsigned>(transfer_.slot.sample_rate),
+      static_cast<unsigned>(transfer_.slot.channels),
+      slot_index,
+      static_cast<unsigned>(free_bytes));
+  return true;
+}
+
+bool StoredFiles::handleData(uint32_t seq, const uint8_t *data, size_t data_len)
+{
+  if (!transfer_.active)
+  {
+    log_w("Stored file data without active transfer");
+    return false;
+  }
+
+  if (seq != transfer_.next_seq)
+  {
+    log_w("Stored file seq gap got=%u expected=%u", static_cast<unsigned>(seq), static_cast<unsigned>(transfer_.next_seq));
+    resetTransfer();
+    return false;
+  }
+  transfer_.next_seq++;
+
+  size_t next_size = transfer_.payload.size() + data_len;
+  if (next_size > transfer_.slot.size || next_size > kMaxStoredFileBytes)
+  {
+    log_w("Stored file payload too large id=%s size=%u expected=%u",
+          transfer_.slot.file_id,
+          static_cast<unsigned>(next_size),
+          static_cast<unsigned>(transfer_.slot.size));
+    resetTransfer();
+    return false;
+  }
+
+  transfer_.payload.insert(transfer_.payload.end(), data, data + data_len);
+  transfer_.received_bytes += static_cast<uint32_t>(data_len);
+  transfer_.chunk_count++;
+  log_i(
+      "Stored file chunk id=%s chunk=%u bytes=%u total=%u/%u",
+      transfer_.slot.file_id,
+      static_cast<unsigned>(transfer_.chunk_count),
+      static_cast<unsigned>(data_len),
+      static_cast<unsigned>(transfer_.received_bytes),
+      static_cast<unsigned>(transfer_.slot.size));
+  return true;
+}
+
+bool StoredFiles::handleEnd(uint32_t seq)
+{
+  if (!transfer_.active)
+  {
+    log_w("Stored file end without active transfer");
+    return false;
+  }
+
+  if (seq != transfer_.next_seq)
+  {
+    log_w("Stored file end seq gap got=%u expected=%u", static_cast<unsigned>(seq), static_cast<unsigned>(transfer_.next_seq));
+    resetTransfer();
+    return false;
+  }
+
+  if (transfer_.payload.size() != transfer_.slot.size)
+  {
+    log_w("Stored file size mismatch id=%s actual=%u expected=%u",
+          transfer_.slot.file_id,
+          static_cast<unsigned>(transfer_.payload.size()),
+          static_cast<unsigned>(transfer_.slot.size));
+    resetTransfer();
+    return false;
+  }
+
+  int slot_index = transfer_.slot_index;
+  PersistedSlot slot = transfer_.slot;
+  std::vector<uint8_t> payload = transfer_.payload;
+
+  if (!persistSlotPayload(slot_index, payload))
+  {
+    resetTransfer();
+    return false;
+  }
+
+  slots_[slot_index] = slot;
+  if (!persistIndex())
+  {
+    resetTransfer();
+    return false;
+  }
+
+  bool activated = activateSlot(slot_index, payload);
+  log_i(
+      "Stored file saved to SPIFFS id=%s slot=%d chunks=%u bytes=%u activated=%u used=%u",
+      slot.file_id,
+      slot_index,
+      static_cast<unsigned>(transfer_.chunk_count),
+      static_cast<unsigned>(payload.size()),
+      static_cast<unsigned>(activated),
+      static_cast<unsigned>(SPIFFS.usedBytes()));
+  resetTransfer();
+  return activated;
+}
+
+bool StoredFiles::getActivePcmFile(const char *fileId, StoredFileView &view)
+{
+  view = StoredFileView{};
+
+  int slot_index = findSlotById(fileId);
+  if (slot_index < 0 || !session_active_[slot_index])
+  {
+    log_i("Stored file inactive or not received in this session id=%s", fileId);
+    return false;
+  }
+
+  const PersistedSlot &slot = slots_[slot_index];
+  if (strcmp(slot.content_type, "audio/pcm") != 0)
+  {
+    log_w("Stored file id=%s has unsupported content_type=%s", slot.file_id, slot.content_type);
+    return false;
+  }
+
+  if (cached_slot_index_ != slot_index)
+  {
+    std::vector<uint8_t> payload;
+    if (!loadSlotPayload(slot_index, payload))
+    {
+      log_w("Failed to load stored file payload id=%s slot=%d", fileId, slot_index);
+      return false;
+    }
+    cached_slot_index_ = slot_index;
+    cached_payload_ = std::move(payload);
+    log_i("Loaded stored file payload from SPIFFS into cache id=%s slot=%d bytes=%u",
+          fileId,
+          slot_index,
+          static_cast<unsigned>(cached_payload_.size()));
+  }
+
+  if (cached_payload_.empty() && slot.size != 0)
+  {
+    return false;
+  }
+
+  view.data = cached_payload_.data();
+  view.size = cached_payload_.size();
+  view.sample_rate = slot.sample_rate;
+  view.channels = static_cast<uint16_t>(slot.channels);
+  log_i("Stored file ready for playback id=%s slot=%d bytes=%u sample_rate=%u channels=%u",
+        fileId,
+        slot_index,
+        static_cast<unsigned>(view.size),
+        static_cast<unsigned>(view.sample_rate),
+        static_cast<unsigned>(view.channels));
+  return true;
+}
+
+bool StoredFiles::mountSpiffs()
+{
+  if (storage_ready_)
+  {
+    return true;
+  }
+
+  storage_ready_ = SPIFFS.begin(true);
+  if (!storage_ready_)
+  {
+    log_w("Failed to mount SPIFFS");
+  }
+  return storage_ready_;
+}
+
+bool StoredFiles::loadIndex()
+{
+  if (!storage_ready_)
+  {
+    return false;
+  }
+
+  if (!SPIFFS.exists(indexPath()))
+  {
+    slots_ = {};
+    return true;
+  }
+
+  File file = SPIFFS.open(indexPath(), "r");
+  if (!file)
+  {
+    log_w("Failed to open SPIFFS index file path=%s", indexPath());
+    return false;
+  }
+
+  size_t index_size = file.size();
+  if (index_size != sizeof(slots_))
+  {
+    log_w("Stored file index size mismatch actual=%u expected=%u",
+          static_cast<unsigned>(index_size),
+          static_cast<unsigned>(sizeof(slots_)));
+    file.close();
+    return false;
+  }
+
+  size_t bytes_read = file.read(reinterpret_cast<uint8_t *>(slots_.data()), sizeof(slots_));
+  file.close();
+  return bytes_read == sizeof(slots_);
+}
+
+bool StoredFiles::persistIndex()
+{
+  if (!storage_ready_)
+  {
+    return false;
+  }
+
+  File file = SPIFFS.open(indexPath(), "w");
+  if (!file)
+  {
+    log_w("Failed to open SPIFFS index file for write path=%s", indexPath());
+    return false;
+  }
+
+  size_t bytes_written = file.write(
+      reinterpret_cast<const uint8_t *>(slots_.data()),
+      sizeof(slots_));
+  file.close();
+  if (bytes_written != sizeof(slots_))
+  {
+    log_w("Failed to persist stored file index written=%u expected=%u",
+          static_cast<unsigned>(bytes_written),
+          static_cast<unsigned>(sizeof(slots_)));
+    return false;
+  }
+  return true;
+}
+
+bool StoredFiles::persistSlotPayload(int slotIndex, const std::vector<uint8_t> &payload)
+{
+  if (!storage_ready_)
+  {
+    return false;
+  }
+
+  File file = SPIFFS.open(payloadPathForSlot(slotIndex), "w");
+  if (!file)
+  {
+    log_w("Failed to open SPIFFS payload file for write slot=%d path=%s",
+          slotIndex,
+          payloadPathForSlot(slotIndex));
+    return false;
+  }
+
+  size_t bytes_written = 0;
+  if (!payload.empty())
+  {
+    bytes_written = file.write(payload.data(), payload.size());
+  }
+  file.close();
+  if (bytes_written != payload.size())
+  {
+    log_w("Failed to persist stored file payload slot=%d written=%u expected=%u",
+          slotIndex,
+          static_cast<unsigned>(bytes_written),
+          static_cast<unsigned>(payload.size()));
+    return false;
+  }
+  return true;
+}
+
+bool StoredFiles::loadSlotPayload(int slotIndex, std::vector<uint8_t> &payload)
+{
+  payload.clear();
+  if (!storage_ready_)
+  {
+    return false;
+  }
+
+  File file = SPIFFS.open(payloadPathForSlot(slotIndex), "r");
+  if (!file)
+  {
+    log_w("Failed to open SPIFFS payload file for read slot=%d path=%s",
+          slotIndex,
+          payloadPathForSlot(slotIndex));
+    return false;
+  }
+
+  size_t payload_size = file.size();
+  if (payload_size != slots_[slotIndex].size)
+  {
+    log_w("Stored file payload size mismatch slot=%d actual=%u expected=%u",
+          slotIndex,
+          static_cast<unsigned>(payload_size),
+          static_cast<unsigned>(slots_[slotIndex].size));
+    file.close();
+    return false;
+  }
+
+  payload.resize(payload_size);
+  size_t bytes_read = 0;
+  if (payload_size > 0)
+  {
+    bytes_read = file.read(payload.data(), payload_size);
+  }
+  file.close();
+  return bytes_read == payload_size;
+}
+
+int StoredFiles::findSlotById(const char *fileId) const
+{
+  for (size_t i = 0; i < slots_.size(); ++i)
+  {
+    if (!slots_[i].used)
+    {
+      continue;
+    }
+    if (strcmp(slots_[i].file_id, fileId) == 0)
+    {
+      return static_cast<int>(i);
+    }
+  }
+  return -1;
+}
+
+int StoredFiles::selectSlotForId(const char *fileId)
+{
+  int existing_slot = findSlotById(fileId);
+  if (existing_slot >= 0)
+  {
+    return existing_slot;
+  }
+
+  for (size_t i = 0; i < slots_.size(); ++i)
+  {
+    if (!slots_[i].used)
+    {
+      return static_cast<int>(i);
+    }
+  }
+
+  return 0;
+}
+
+void StoredFiles::resetTransfer()
+{
+  transfer_ = TransferState{};
+}
+
+bool StoredFiles::activateSlot(int slotIndex, const std::vector<uint8_t> &payload)
+{
+  if (slotIndex < 0 || slotIndex >= static_cast<int>(slots_.size()))
+  {
+    return false;
+  }
+
+  session_active_[slotIndex] = true;
+  cached_slot_index_ = slotIndex;
+  cached_payload_ = payload;
+  log_i("Stored file activated for current session id=%s slot=%d bytes=%u",
+        slots_[slotIndex].file_id,
+        slotIndex,
+        static_cast<unsigned>(cached_payload_.size()));
+  return true;
+}
+
+const char *StoredFiles::payloadPathForSlot(int slotIndex)
+{
+  if (slotIndex < 0 || slotIndex >= static_cast<int>(sizeof(kPayloadPaths) / sizeof(kPayloadPaths[0])))
+  {
+    return kPayloadPaths[0];
+  }
+  return kPayloadPaths[slotIndex];
+}
+
+const char *StoredFiles::indexPath()
+{
+  return kIndexFilePath;
+}
diff --git a/misc/wake_sound/wake_sound.wav b/misc/wake_sound/wake_sound.wav
new file mode 100644
index 0000000..a1f37c5
Binary files /dev/null and b/misc/wake_sound/wake_sound.wav differ
diff --git a/protobuf/websocket-message.options b/protobuf/websocket-message.options
index e032f4f..ab152ff 100644
--- a/protobuf/websocket-message.options
+++ b/protobuf/websocket-message.options
@@ -1,4 +1,7 @@
 stackchan.websocket.v1.AudioChunk.pcm_bytes max_size:4096
+stackchan.websocket.v1.FileChunk.chunk_bytes max_size:4096
 stackchan.websocket.v1.ServoCommandSequence.commands max_count:255
 stackchan.websocket.v1.FirmwareMetadata.firmware_version max_length:63
 stackchan.websocket.v1.ServerMetadata.server_version max_length:63
+stackchan.websocket.v1.StoredFileStart.file_id max_length:63
+stackchan.websocket.v1.StoredFileStart.content_type max_length:63
diff --git a/protobuf/websocket-message.proto b/protobuf/websocket-message.proto
index c643673..daae0c8 100644
--- a/protobuf/websocket-message.proto
+++ b/protobuf/websocket-message.proto
@@ -31,6 +31,9 @@ message WebSocketMessage {
     ServoDoneEvent servo_done_evt = 35;
     FirmwareMetadata firmware_metadata = 36;
     ServerMetadata server_metadata = 37;
+    StoredFileStart stored_file_start = 40;
+    FileChunk stored_file_data = 41;
+    StoredFileEnd stored_file_end = 42;
   }
 }
 
@@ -46,6 +49,7 @@ enum MessageKind {
   MESSAGE_KIND_SERVO_DONE_EVT = 8;
   MESSAGE_KIND_FIRMWARE_METADATA = 9;
   MESSAGE_KIND_SERVER_METADATA = 10;
+  MESSAGE_KIND_STORED_FILE = 11;
 }
 
 enum MessageType {
@@ -97,6 +101,10 @@ message AudioChunk {
   bytes pcm_bytes = 1;
 }
 
+message FileChunk {
+  bytes chunk_bytes = 1;
+}
+
 message StateCommand {
   StackchanState state = 1;
 }
@@ -113,6 +121,16 @@ message SpeakDoneEvent {
   bool done = 1;
 }
 
+message StoredFileStart {
+  string file_id = 1;
+  string content_type = 2;
+  uint32 total_size = 3;
+  uint32 sample_rate = 4;
+  uint32 channels = 5;
+}
+
+message StoredFileEnd {}
+
 message ServoCommandSequence {
   repeated ServoCommand commands = 1;
 }
diff --git a/stackchan_server/app.py b/stackchan_server/app.py
index 14496d2..3ca50f8 100644
--- a/stackchan_server/app.py
+++ b/stackchan_server/app.py
@@ -29,9 +29,12 @@ def __init__(
         self,
         speech_recognizer: SpeechRecognizer | None = None,
         speech_synthesizer: SpeechSynthesizer | None = None,
+        *,
+        send_wakeword_sound_on_connect: bool = True,
     ) -> None:
         self.speech_recognizer = speech_recognizer or create_speech_recognizer()
         self.speech_synthesizer = speech_synthesizer or create_speech_synthesizer()
+        self.send_wakeword_sound_on_connect = send_wakeword_sound_on_connect
         self.fastapi = FastAPI(title="StackChan WebSocket Server")
         self._setup_fn: Optional[Callable[[WsProxy], Awaitable[None]]] = None
         self._talk_session_fn: Optional[Callable[[WsProxy], Awaitable[None]]] = None
@@ -96,6 +99,9 @@ async def _handle_ws(self, websocket: WebSocket) -> None:
             await existing.close()
 
         try:
+            if self.send_wakeword_sound_on_connect:
+                await proxy.send_default_wake_word_sound()
+
             if self._setup_fn:
                 await self._setup_fn(proxy)
 
diff --git a/stackchan_server/generated_protobuf/websocket_message_pb2.py b/stackchan_server/generated_protobuf/websocket_message_pb2.py
index a7d7a4e..9d4a505 100644
--- a/stackchan_server/generated_protobuf/websocket_message_pb2.py
+++ b/stackchan_server/generated_protobuf/websocket_message_pb2.py
@@ -24,53 +24,59 @@
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x17websocket-message.proto\x12\x16stackchan.websocket.v1\"\x96\x08\n\x10WebSocketMessage\x12\x31\n\x04kind\x18\x01 \x01(\x0e\x32#.stackchan.websocket.v1.MessageKind\x12\x39\n\x0cmessage_type\x18\x02 \x01(\x0e\x32#.stackchan.websocket.v1.MessageType\x12\x0b\n\x03seq\x18\x03 \x01(\r\x12@\n\x0f\x61udio_pcm_start\x18\n \x01(\x0b\x32%.stackchan.websocket.v1.AudioPcmStartH\x00\x12<\n\x0e\x61udio_pcm_data\x18\x0b \x01(\x0b\x32\".stackchan.websocket.v1.AudioChunkH\x00\x12<\n\raudio_pcm_end\x18\x0c \x01(\x0b\x32#.stackchan.websocket.v1.AudioPcmEndH\x00\x12@\n\x0f\x61udio_wav_start\x18\x14 \x01(\x0b\x32%.stackchan.websocket.v1.AudioWavStartH\x00\x12<\n\x0e\x61udio_wav_data\x18\x15 \x01(\x0b\x32\".stackchan.websocket.v1.AudioChunkH\x00\x12<\n\raudio_wav_end\x18\x16 \x01(\x0b\x32#.stackchan.websocket.v1.AudioWavEndH\x00\x12\x39\n\tstate_cmd\x18\x1e \x01(\x0b\x32$.stackchan.websocket.v1.StateCommandH\x00\x12>\n\rwake_word_evt\x18\x1f \x01(\x0b\x32%.stackchan.websocket.v1.WakeWordEventH\x00\x12\x37\n\tstate_evt\x18  \x01(\x0b\x32\".stackchan.websocket.v1.StateEventH\x00\x12@\n\x0espeak_done_evt\x18! \x01(\x0b\x32&.stackchan.websocket.v1.SpeakDoneEventH\x00\x12\x41\n\tservo_cmd\x18\" \x01(\x0b\x32,.stackchan.websocket.v1.ServoCommandSequenceH\x00\x12@\n\x0eservo_done_evt\x18# \x01(\x0b\x32&.stackchan.websocket.v1.ServoDoneEventH\x00\x12\x45\n\x11\x66irmware_metadata\x18$ \x01(\x0b\x32(.stackchan.websocket.v1.FirmwareMetadataH\x00\x12\x41\n\x0fserver_metadata\x18% \x01(\x0b\x32&.stackchan.websocket.v1.ServerMetadataH\x00\x42\x06\n\x04\x62ody\"\x0f\n\rAudioPcmStart\"\r\n\x0b\x41udioPcmEnd\"6\n\rAudioWavStart\x12\x13\n\x0bsample_rate\x18\x01 \x01(\r\x12\x10\n\x08\x63hannels\x18\x02 \x01(\r\"\r\n\x0b\x41udioWavEnd\"\x1f\n\nAudioChunk\x12\x11\n\tpcm_bytes\x18\x01 \x01(\x0c\"E\n\x0cStateCommand\x12\x35\n\x05state\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.StackchanState\"!\n\rWakeWordEvent\x12\x10\n\x08\x64\x65tected\x18\x01 \x01(\x08\"C\n\nStateEvent\x12\x35\n\x05state\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.StackchanState\"\x1e\n\x0eSpeakDoneEvent\x12\x0c\n\x04\x64one\x18\x01 \x01(\x08\"N\n\x14ServoCommandSequence\x12\x36\n\x08\x63ommands\x18\x01 \x03(\x0b\x32$.stackchan.websocket.v1.ServoCommand\"f\n\x0cServoCommand\x12\x32\n\x02op\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.ServoOperation\x12\r\n\x05\x61ngle\x18\x02 \x01(\x11\x12\x13\n\x0b\x64uration_ms\x18\x03 \x01(\x11\"\x1e\n\x0eServoDoneEvent\x12\x0c\n\x04\x64one\x18\x01 \x01(\x08\"\x99\x02\n\x10\x46irmwareMetadata\x12\x37\n\x0b\x64\x65vice_type\x18\x01 \x01(\x0e\x32\".stackchan.websocket.v1.DeviceType\x12\x15\n\rdisplay_width\x18\x02 \x01(\r\x12\x16\n\x0e\x64isplay_height\x18\x03 \x01(\r\x12\x1c\n\x14has_device_wake_word\x18\x04 \x01(\x08\x12\x0f\n\x07has_led\x18\x05 \x01(\x08\x12\x35\n\nservo_type\x18\x06 \x01(\x0e\x32!.stackchan.websocket.v1.ServoType\x12\x1d\n\x15supports_audio_duplex\x18\x07 \x01(\x08\x12\x18\n\x10\x66irmware_version\x18\x08 \x01(\t\"F\n\x0eServerMetadata\x12\x1c\n\x14has_server_wake_word\x18\x01 \x01(\x08\x12\x16\n\x0eserver_version\x18\x02 \x01(\t*\xdf\x02\n\x0bMessageKind\x12\x1c\n\x18MESSAGE_KIND_UNSPECIFIED\x10\x00\x12\x1a\n\x16MESSAGE_KIND_AUDIO_PCM\x10\x01\x12\x1a\n\x16MESSAGE_KIND_AUDIO_WAV\x10\x02\x12\x1a\n\x16MESSAGE_KIND_STATE_CMD\x10\x03\x12\x1e\n\x1aMESSAGE_KIND_WAKE_WORD_EVT\x10\x04\x12\x1a\n\x16MESSAGE_KIND_STATE_EVT\x10\x05\x12\x1f\n\x1bMESSAGE_KIND_SPEAK_DONE_EVT\x10\x06\x12\x1a\n\x16MESSAGE_KIND_SERVO_CMD\x10\x07\x12\x1f\n\x1bMESSAGE_KIND_SERVO_DONE_EVT\x10\x08\x12\"\n\x1eMESSAGE_KIND_FIRMWARE_METADATA\x10\t\x12 \n\x1cMESSAGE_KIND_SERVER_METADATA\x10\n*p\n\x0bMessageType\x12\x1c\n\x18MESSAGE_TYPE_UNSPECIFIED\x10\x00\x12\x16\n\x12MESSAGE_TYPE_START\x10\x01\x12\x15\n\x11MESSAGE_TYPE_DATA\x10\x02\x12\x14\n\x10MESSAGE_TYPE_END\x10\x03*\x85\x01\n\x0eStackchanState\x12\x18\n\x14STACKCHAN_STATE_IDLE\x10\x00\x12\x1d\n\x19STACKCHAN_STATE_LISTENING\x10\x01\x12\x1c\n\x18STACKCHAN_STATE_THINKING\x10\x02\x12\x1c\n\x18STACKCHAN_STATE_SPEAKING\x10\x03*c\n\x0eServoOperation\x12\x19\n\x15SERVO_OPERATION_SLEEP\x10\x00\x12\x1a\n\x16SERVO_OPERATION_MOVE_X\x10\x01\x12\x1a\n\x16SERVO_OPERATION_MOVE_Y\x10\x02*\x85\x01\n\nDeviceType\x12\x1b\n\x17\x44\x45VICE_TYPE_UNSPECIFIED\x10\x00\x12\x1e\n\x1a\x44\x45VICE_TYPE_M5STACK_CORES3\x10\x01\x12\x1a\n\x16\x44\x45VICE_TYPE_M5ATOM_S3R\x10\x02\x12\x1e\n\x1a\x44\x45VICE_TYPE_M5ATOM_ECHOS3R\x10\x03*i\n\tServoType\x12\x1a\n\x16SERVO_TYPE_UNSPECIFIED\x10\x00\x12\x13\n\x0fSERVO_TYPE_NONE\x10\x01\x12\x13\n\x0fSERVO_TYPE_SG90\x10\x02\x12\x16\n\x12SERVO_TYPE_SCS0009\x10\x03\x62\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x17websocket-message.proto\x12\x16stackchan.websocket.v1\"\xdd\t\n\x10WebSocketMessage\x12\x31\n\x04kind\x18\x01 \x01(\x0e\x32#.stackchan.websocket.v1.MessageKind\x12\x39\n\x0cmessage_type\x18\x02 \x01(\x0e\x32#.stackchan.websocket.v1.MessageType\x12\x0b\n\x03seq\x18\x03 \x01(\r\x12@\n\x0f\x61udio_pcm_start\x18\n \x01(\x0b\x32%.stackchan.websocket.v1.AudioPcmStartH\x00\x12<\n\x0e\x61udio_pcm_data\x18\x0b \x01(\x0b\x32\".stackchan.websocket.v1.AudioChunkH\x00\x12<\n\raudio_pcm_end\x18\x0c \x01(\x0b\x32#.stackchan.websocket.v1.AudioPcmEndH\x00\x12@\n\x0f\x61udio_wav_start\x18\x14 \x01(\x0b\x32%.stackchan.websocket.v1.AudioWavStartH\x00\x12<\n\x0e\x61udio_wav_data\x18\x15 \x01(\x0b\x32\".stackchan.websocket.v1.AudioChunkH\x00\x12<\n\raudio_wav_end\x18\x16 \x01(\x0b\x32#.stackchan.websocket.v1.AudioWavEndH\x00\x12\x39\n\tstate_cmd\x18\x1e \x01(\x0b\x32$.stackchan.websocket.v1.StateCommandH\x00\x12>\n\rwake_word_evt\x18\x1f \x01(\x0b\x32%.stackchan.websocket.v1.WakeWordEventH\x00\x12\x37\n\tstate_evt\x18  \x01(\x0b\x32\".stackchan.websocket.v1.StateEventH\x00\x12@\n\x0espeak_done_evt\x18! \x01(\x0b\x32&.stackchan.websocket.v1.SpeakDoneEventH\x00\x12\x41\n\tservo_cmd\x18\" \x01(\x0b\x32,.stackchan.websocket.v1.ServoCommandSequenceH\x00\x12@\n\x0eservo_done_evt\x18# \x01(\x0b\x32&.stackchan.websocket.v1.ServoDoneEventH\x00\x12\x45\n\x11\x66irmware_metadata\x18$ \x01(\x0b\x32(.stackchan.websocket.v1.FirmwareMetadataH\x00\x12\x41\n\x0fserver_metadata\x18% \x01(\x0b\x32&.stackchan.websocket.v1.ServerMetadataH\x00\x12\x44\n\x11stored_file_start\x18( \x01(\x0b\x32\'.stackchan.websocket.v1.StoredFileStartH\x00\x12=\n\x10stored_file_data\x18) \x01(\x0b\x32!.stackchan.websocket.v1.FileChunkH\x00\x12@\n\x0fstored_file_end\x18* \x01(\x0b\x32%.stackchan.websocket.v1.StoredFileEndH\x00\x42\x06\n\x04\x62ody\"\x0f\n\rAudioPcmStart\"\r\n\x0b\x41udioPcmEnd\"6\n\rAudioWavStart\x12\x13\n\x0bsample_rate\x18\x01 \x01(\r\x12\x10\n\x08\x63hannels\x18\x02 \x01(\r\"\r\n\x0b\x41udioWavEnd\"\x1f\n\nAudioChunk\x12\x11\n\tpcm_bytes\x18\x01 \x01(\x0c\" \n\tFileChunk\x12\x13\n\x0b\x63hunk_bytes\x18\x01 \x01(\x0c\"E\n\x0cStateCommand\x12\x35\n\x05state\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.StackchanState\"!\n\rWakeWordEvent\x12\x10\n\x08\x64\x65tected\x18\x01 \x01(\x08\"C\n\nStateEvent\x12\x35\n\x05state\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.StackchanState\"\x1e\n\x0eSpeakDoneEvent\x12\x0c\n\x04\x64one\x18\x01 \x01(\x08\"s\n\x0fStoredFileStart\x12\x0f\n\x07\x66ile_id\x18\x01 \x01(\t\x12\x14\n\x0c\x63ontent_type\x18\x02 \x01(\t\x12\x12\n\ntotal_size\x18\x03 \x01(\r\x12\x13\n\x0bsample_rate\x18\x04 \x01(\r\x12\x10\n\x08\x63hannels\x18\x05 \x01(\r\"\x0f\n\rStoredFileEnd\"N\n\x14ServoCommandSequence\x12\x36\n\x08\x63ommands\x18\x01 \x03(\x0b\x32$.stackchan.websocket.v1.ServoCommand\"f\n\x0cServoCommand\x12\x32\n\x02op\x18\x01 \x01(\x0e\x32&.stackchan.websocket.v1.ServoOperation\x12\r\n\x05\x61ngle\x18\x02 \x01(\x11\x12\x13\n\x0b\x64uration_ms\x18\x03 \x01(\x11\"\x1e\n\x0eServoDoneEvent\x12\x0c\n\x04\x64one\x18\x01 \x01(\x08\"\x99\x02\n\x10\x46irmwareMetadata\x12\x37\n\x0b\x64\x65vice_type\x18\x01 \x01(\x0e\x32\".stackchan.websocket.v1.DeviceType\x12\x15\n\rdisplay_width\x18\x02 \x01(\r\x12\x16\n\x0e\x64isplay_height\x18\x03 \x01(\r\x12\x1c\n\x14has_device_wake_word\x18\x04 \x01(\x08\x12\x0f\n\x07has_led\x18\x05 \x01(\x08\x12\x35\n\nservo_type\x18\x06 \x01(\x0e\x32!.stackchan.websocket.v1.ServoType\x12\x1d\n\x15supports_audio_duplex\x18\x07 \x01(\x08\x12\x18\n\x10\x66irmware_version\x18\x08 \x01(\t\"F\n\x0eServerMetadata\x12\x1c\n\x14has_server_wake_word\x18\x01 \x01(\x08\x12\x16\n\x0eserver_version\x18\x02 \x01(\t*\xfd\x02\n\x0bMessageKind\x12\x1c\n\x18MESSAGE_KIND_UNSPECIFIED\x10\x00\x12\x1a\n\x16MESSAGE_KIND_AUDIO_PCM\x10\x01\x12\x1a\n\x16MESSAGE_KIND_AUDIO_WAV\x10\x02\x12\x1a\n\x16MESSAGE_KIND_STATE_CMD\x10\x03\x12\x1e\n\x1aMESSAGE_KIND_WAKE_WORD_EVT\x10\x04\x12\x1a\n\x16MESSAGE_KIND_STATE_EVT\x10\x05\x12\x1f\n\x1bMESSAGE_KIND_SPEAK_DONE_EVT\x10\x06\x12\x1a\n\x16MESSAGE_KIND_SERVO_CMD\x10\x07\x12\x1f\n\x1bMESSAGE_KIND_SERVO_DONE_EVT\x10\x08\x12\"\n\x1eMESSAGE_KIND_FIRMWARE_METADATA\x10\t\x12 \n\x1cMESSAGE_KIND_SERVER_METADATA\x10\n\x12\x1c\n\x18MESSAGE_KIND_STORED_FILE\x10\x0b*p\n\x0bMessageType\x12\x1c\n\x18MESSAGE_TYPE_UNSPECIFIED\x10\x00\x12\x16\n\x12MESSAGE_TYPE_START\x10\x01\x12\x15\n\x11MESSAGE_TYPE_DATA\x10\x02\x12\x14\n\x10MESSAGE_TYPE_END\x10\x03*\x85\x01\n\x0eStackchanState\x12\x18\n\x14STACKCHAN_STATE_IDLE\x10\x00\x12\x1d\n\x19STACKCHAN_STATE_LISTENING\x10\x01\x12\x1c\n\x18STACKCHAN_STATE_THINKING\x10\x02\x12\x1c\n\x18STACKCHAN_STATE_SPEAKING\x10\x03*c\n\x0eServoOperation\x12\x19\n\x15SERVO_OPERATION_SLEEP\x10\x00\x12\x1a\n\x16SERVO_OPERATION_MOVE_X\x10\x01\x12\x1a\n\x16SERVO_OPERATION_MOVE_Y\x10\x02*\x85\x01\n\nDeviceType\x12\x1b\n\x17\x44\x45VICE_TYPE_UNSPECIFIED\x10\x00\x12\x1e\n\x1a\x44\x45VICE_TYPE_M5STACK_CORES3\x10\x01\x12\x1a\n\x16\x44\x45VICE_TYPE_M5ATOM_S3R\x10\x02\x12\x1e\n\x1a\x44\x45VICE_TYPE_M5ATOM_ECHOS3R\x10\x03*i\n\tServoType\x12\x1a\n\x16SERVO_TYPE_UNSPECIFIED\x10\x00\x12\x13\n\x0fSERVO_TYPE_NONE\x10\x01\x12\x13\n\x0fSERVO_TYPE_SG90\x10\x02\x12\x16\n\x12SERVO_TYPE_SCS0009\x10\x03\x62\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)
 _builder.BuildTopDescriptorsAndMessages(DESCRIPTOR, 'websocket_message_pb2', _globals)
 if not _descriptor._USE_C_DESCRIPTORS:
   DESCRIPTOR._loaded_options = None
-  _globals['_MESSAGEKIND']._serialized_start=2016
-  _globals['_MESSAGEKIND']._serialized_end=2367
-  _globals['_MESSAGETYPE']._serialized_start=2369
-  _globals['_MESSAGETYPE']._serialized_end=2481
-  _globals['_STACKCHANSTATE']._serialized_start=2484
-  _globals['_STACKCHANSTATE']._serialized_end=2617
-  _globals['_SERVOOPERATION']._serialized_start=2619
-  _globals['_SERVOOPERATION']._serialized_end=2718
-  _globals['_DEVICETYPE']._serialized_start=2721
-  _globals['_DEVICETYPE']._serialized_end=2854
-  _globals['_SERVOTYPE']._serialized_start=2856
-  _globals['_SERVOTYPE']._serialized_end=2961
+  _globals['_MESSAGEKIND']._serialized_start=2383
+  _globals['_MESSAGEKIND']._serialized_end=2764
+  _globals['_MESSAGETYPE']._serialized_start=2766
+  _globals['_MESSAGETYPE']._serialized_end=2878
+  _globals['_STACKCHANSTATE']._serialized_start=2881
+  _globals['_STACKCHANSTATE']._serialized_end=3014
+  _globals['_SERVOOPERATION']._serialized_start=3016
+  _globals['_SERVOOPERATION']._serialized_end=3115
+  _globals['_DEVICETYPE']._serialized_start=3118
+  _globals['_DEVICETYPE']._serialized_end=3251
+  _globals['_SERVOTYPE']._serialized_start=3253
+  _globals['_SERVOTYPE']._serialized_end=3358
   _globals['_WEBSOCKETMESSAGE']._serialized_start=52
-  _globals['_WEBSOCKETMESSAGE']._serialized_end=1098
-  _globals['_AUDIOPCMSTART']._serialized_start=1100
-  _globals['_AUDIOPCMSTART']._serialized_end=1115
-  _globals['_AUDIOPCMEND']._serialized_start=1117
-  _globals['_AUDIOPCMEND']._serialized_end=1130
-  _globals['_AUDIOWAVSTART']._serialized_start=1132
-  _globals['_AUDIOWAVSTART']._serialized_end=1186
-  _globals['_AUDIOWAVEND']._serialized_start=1188
-  _globals['_AUDIOWAVEND']._serialized_end=1201
-  _globals['_AUDIOCHUNK']._serialized_start=1203
-  _globals['_AUDIOCHUNK']._serialized_end=1234
-  _globals['_STATECOMMAND']._serialized_start=1236
-  _globals['_STATECOMMAND']._serialized_end=1305
-  _globals['_WAKEWORDEVENT']._serialized_start=1307
-  _globals['_WAKEWORDEVENT']._serialized_end=1340
-  _globals['_STATEEVENT']._serialized_start=1342
-  _globals['_STATEEVENT']._serialized_end=1409
-  _globals['_SPEAKDONEEVENT']._serialized_start=1411
-  _globals['_SPEAKDONEEVENT']._serialized_end=1441
-  _globals['_SERVOCOMMANDSEQUENCE']._serialized_start=1443
-  _globals['_SERVOCOMMANDSEQUENCE']._serialized_end=1521
-  _globals['_SERVOCOMMAND']._serialized_start=1523
-  _globals['_SERVOCOMMAND']._serialized_end=1625
-  _globals['_SERVODONEEVENT']._serialized_start=1627
-  _globals['_SERVODONEEVENT']._serialized_end=1657
-  _globals['_FIRMWAREMETADATA']._serialized_start=1660
-  _globals['_FIRMWAREMETADATA']._serialized_end=1941
-  _globals['_SERVERMETADATA']._serialized_start=1943
-  _globals['_SERVERMETADATA']._serialized_end=2013
+  _globals['_WEBSOCKETMESSAGE']._serialized_end=1297
+  _globals['_AUDIOPCMSTART']._serialized_start=1299
+  _globals['_AUDIOPCMSTART']._serialized_end=1314
+  _globals['_AUDIOPCMEND']._serialized_start=1316
+  _globals['_AUDIOPCMEND']._serialized_end=1329
+  _globals['_AUDIOWAVSTART']._serialized_start=1331
+  _globals['_AUDIOWAVSTART']._serialized_end=1385
+  _globals['_AUDIOWAVEND']._serialized_start=1387
+  _globals['_AUDIOWAVEND']._serialized_end=1400
+  _globals['_AUDIOCHUNK']._serialized_start=1402
+  _globals['_AUDIOCHUNK']._serialized_end=1433
+  _globals['_FILECHUNK']._serialized_start=1435
+  _globals['_FILECHUNK']._serialized_end=1467
+  _globals['_STATECOMMAND']._serialized_start=1469
+  _globals['_STATECOMMAND']._serialized_end=1538
+  _globals['_WAKEWORDEVENT']._serialized_start=1540
+  _globals['_WAKEWORDEVENT']._serialized_end=1573
+  _globals['_STATEEVENT']._serialized_start=1575
+  _globals['_STATEEVENT']._serialized_end=1642
+  _globals['_SPEAKDONEEVENT']._serialized_start=1644
+  _globals['_SPEAKDONEEVENT']._serialized_end=1674
+  _globals['_STOREDFILESTART']._serialized_start=1676
+  _globals['_STOREDFILESTART']._serialized_end=1791
+  _globals['_STOREDFILEEND']._serialized_start=1793
+  _globals['_STOREDFILEEND']._serialized_end=1808
+  _globals['_SERVOCOMMANDSEQUENCE']._serialized_start=1810
+  _globals['_SERVOCOMMANDSEQUENCE']._serialized_end=1888
+  _globals['_SERVOCOMMAND']._serialized_start=1890
+  _globals['_SERVOCOMMAND']._serialized_end=1992
+  _globals['_SERVODONEEVENT']._serialized_start=1994
+  _globals['_SERVODONEEVENT']._serialized_end=2024
+  _globals['_FIRMWAREMETADATA']._serialized_start=2027
+  _globals['_FIRMWAREMETADATA']._serialized_end=2308
+  _globals['_SERVERMETADATA']._serialized_start=2310
+  _globals['_SERVERMETADATA']._serialized_end=2380
 # @@protoc_insertion_point(module_scope)
diff --git a/stackchan_server/protobuf_ws.py b/stackchan_server/protobuf_ws.py
index 8569004..e84d2d1 100644
--- a/stackchan_server/protobuf_ws.py
+++ b/stackchan_server/protobuf_ws.py
@@ -92,6 +92,48 @@ def encode_audio_wav_end_message(seq: int) -> bytes:
     return message.SerializeToString()
 
 
+def encode_stored_file_start_message(
+    seq: int,
+    *,
+    file_id: str,
+    content_type: str,
+    total_size: int,
+    sample_rate: int = 0,
+    channels: int = 0,
+) -> bytes:
+    message = _new_message(
+        ws_pb2.MESSAGE_KIND_STORED_FILE,
+        ws_pb2.MESSAGE_TYPE_START,
+        seq,
+    )
+    message.stored_file_start.file_id = file_id
+    message.stored_file_start.content_type = content_type
+    message.stored_file_start.total_size = int(total_size)
+    message.stored_file_start.sample_rate = int(sample_rate)
+    message.stored_file_start.channels = int(channels)
+    return message.SerializeToString()
+
+
+def encode_stored_file_data_message(seq: int, chunk_bytes: bytes) -> bytes:
+    message = _new_message(
+        ws_pb2.MESSAGE_KIND_STORED_FILE,
+        ws_pb2.MESSAGE_TYPE_DATA,
+        seq,
+    )
+    message.stored_file_data.chunk_bytes = chunk_bytes
+    return message.SerializeToString()
+
+
+def encode_stored_file_end_message(seq: int) -> bytes:
+    message = _new_message(
+        ws_pb2.MESSAGE_KIND_STORED_FILE,
+        ws_pb2.MESSAGE_TYPE_END,
+        seq,
+    )
+    message.stored_file_end.SetInParent()
+    return message.SerializeToString()
+
+
 def encode_state_command_message(seq: int, state_id: int) -> bytes:
     message = _new_message(
         ws_pb2.MESSAGE_KIND_STATE_CMD,
@@ -181,6 +223,9 @@ def encode_servo_command_message(seq: int, commands: Sequence[ServoCommand]) ->
     "encode_audio_wav_start_message",
     "encode_server_metadata_message",
     "encode_servo_command_message",
+    "encode_stored_file_data_message",
+    "encode_stored_file_end_message",
+    "encode_stored_file_start_message",
     "encode_state_command_message",
     "parse_websocket_message",
     "ws_pb2",
diff --git a/stackchan_server/wakeword_sound.py b/stackchan_server/wakeword_sound.py
new file mode 100644
index 0000000..e746f0c
--- /dev/null
+++ b/stackchan_server/wakeword_sound.py
@@ -0,0 +1,201 @@
+from __future__ import annotations
+
+import os
+import sys
+import wave
+from array import array
+from dataclasses import dataclass
+from logging import getLogger
+from pathlib import Path
+
+DEFAULT_WAKE_WORD_SOUND_FILE_ID = "wakeword-detected-sound"
+DEFAULT_WAKE_WORD_SOUND_CONTENT_TYPE = "audio/pcm"
+WAKEWORD_SOUND_PATH_ENV_VAR = "STACKCHAN_WAKEWORD_SOUND_PATH"
+WAKEWORD_SOUND_TARGET_SAMPLE_RATE = 24000
+WAKEWORD_SOUND_TARGET_CHANNELS = 1
+WAKEWORD_SOUND_PREROLL_MS = 40
+WAKEWORD_SOUND_POSTROLL_MS = 180
+WAKEWORD_SOUND_MIN_DURATION_MS = 700
+
+logger = getLogger(__name__)
+
+
+@dataclass(frozen=True)
+class WakeWordSound:
+    file_id: str
+    content_type: str
+    payload: bytes
+    sample_rate: int
+    channels: int
+
+
+def _decode_pcm16le(payload: bytes) -> array[int]:
+    samples = array("h")
+    samples.frombytes(payload)
+    if sys.byteorder != "little":
+        samples.byteswap()
+    return samples
+
+
+def _encode_pcm16le(samples: array[int]) -> bytes:
+    encoded = array("h", samples)
+    if sys.byteorder != "little":
+        encoded.byteswap()
+    return encoded.tobytes()
+
+
+def _clamp_pcm16(value: float) -> int:
+    return max(-32768, min(32767, int(round(value))))
+
+
+def _pcm16_silence(sample_count: int) -> array[int]:
+    if sample_count <= 0:
+        return array("h")
+    return array("h", [0]) * sample_count
+
+
+def _mix_to_mono(samples: array[int], channels: int) -> array[int]:
+    if channels <= 1:
+        return array("h", samples)
+
+    mono_samples = array("h")
+    frame_count = len(samples) // channels
+    for frame_index in range(frame_count):
+        base = frame_index * channels
+        mixed = sum(samples[base + channel_index] for channel_index in range(channels)) / channels
+        mono_samples.append(_clamp_pcm16(mixed))
+    return mono_samples
+
+
+def _resample_mono_pcm16(samples: array[int], src_rate: int, dst_rate: int) -> array[int]:
+    if src_rate == dst_rate or len(samples) <= 1:
+        return array("h", samples)
+
+    dst_length = max(1, round(len(samples) * dst_rate / src_rate))
+    resampled = array("h")
+    for dst_index in range(dst_length):
+        src_position = dst_index * src_rate / dst_rate
+        left_index = int(src_position)
+        if left_index >= len(samples) - 1:
+            resampled.append(samples[-1])
+            continue
+
+        right_index = left_index + 1
+        fraction = src_position - left_index
+        interpolated = samples[left_index] + (samples[right_index] - samples[left_index]) * fraction
+        resampled.append(_clamp_pcm16(interpolated))
+    return resampled
+
+
+def _normalize_pcm16_payload(payload: bytes, sample_rate: int, channels: int) -> tuple[bytes, int, int]:
+    samples = _decode_pcm16le(payload)
+    mono_samples = _mix_to_mono(samples, channels)
+    normalized_samples = _resample_mono_pcm16(
+        mono_samples,
+        sample_rate,
+        WAKEWORD_SOUND_TARGET_SAMPLE_RATE,
+    )
+    return (
+        _encode_pcm16le(normalized_samples),
+        WAKEWORD_SOUND_TARGET_SAMPLE_RATE,
+        WAKEWORD_SOUND_TARGET_CHANNELS,
+    )
+
+
+def _pad_pcm16_for_short_playback(payload: bytes, sample_rate: int) -> bytes:
+    samples = _decode_pcm16le(payload)
+    preroll_samples = round(sample_rate * WAKEWORD_SOUND_PREROLL_MS / 1000)
+    postroll_samples = round(sample_rate * WAKEWORD_SOUND_POSTROLL_MS / 1000)
+    min_duration_samples = round(sample_rate * WAKEWORD_SOUND_MIN_DURATION_MS / 1000)
+
+    padded = _pcm16_silence(preroll_samples)
+    padded.extend(samples)
+    padded.extend(_pcm16_silence(postroll_samples))
+
+    if len(padded) < min_duration_samples:
+        padded.extend(_pcm16_silence(min_duration_samples - len(padded)))
+
+    return _encode_pcm16le(padded)
+
+
+def load_wake_word_detected_sound_from_env() -> WakeWordSound | None:
+    raw_path = os.getenv(WAKEWORD_SOUND_PATH_ENV_VAR, "").strip()
+    if not raw_path:
+        logger.info(
+            "Wake-word sound WAV path is not configured: env=%s",
+            WAKEWORD_SOUND_PATH_ENV_VAR,
+        )
+        return None
+
+    wav_path = Path(raw_path).expanduser()
+    if not wav_path.is_absolute():
+        wav_path = Path.cwd() / wav_path
+
+    resolved_path = wav_path.resolve()
+    logger.info("Loading wake-word sound WAV from %s", resolved_path)
+    if not resolved_path.is_file():
+        raise FileNotFoundError(
+            f"wake word sound wav file not found: {resolved_path}"
+        )
+
+    with wave.open(str(resolved_path), "rb") as wav_fp:
+        channels = wav_fp.getnchannels()
+        sample_width = wav_fp.getsampwidth()
+        sample_rate = wav_fp.getframerate()
+        payload = wav_fp.readframes(wav_fp.getnframes())
+
+    if sample_width != 2:
+        raise ValueError(
+            "wake word notification sound wav must be 16-bit PCM"
+        )
+    if sample_rate <= 0:
+        raise ValueError("wake word notification sound wav has invalid sample rate")
+    if channels <= 0:
+        raise ValueError("wake word notification sound wav has invalid channels")
+    if not payload:
+        raise ValueError("wake word notification sound wav is empty")
+
+    normalized_payload, normalized_sample_rate, normalized_channels = _normalize_pcm16_payload(
+        payload,
+        sample_rate,
+        channels,
+    )
+    playback_ready_payload = _pad_pcm16_for_short_playback(
+        normalized_payload,
+        normalized_sample_rate,
+    )
+
+    logger.info(
+        "Loaded wake-word sound WAV path=%s source_sample_rate=%d source_channels=%d source_bytes=%d normalized_sample_rate=%d normalized_channels=%d normalized_bytes=%d playback_ready_bytes=%d preroll_ms=%d postroll_ms=%d min_duration_ms=%d",
+        resolved_path,
+        sample_rate,
+        channels,
+        len(payload),
+        normalized_sample_rate,
+        normalized_channels,
+        len(normalized_payload),
+        len(playback_ready_payload),
+        WAKEWORD_SOUND_PREROLL_MS,
+        WAKEWORD_SOUND_POSTROLL_MS,
+        WAKEWORD_SOUND_MIN_DURATION_MS,
+    )
+
+    return WakeWordSound(
+        file_id=DEFAULT_WAKE_WORD_SOUND_FILE_ID,
+        content_type=DEFAULT_WAKE_WORD_SOUND_CONTENT_TYPE,
+        payload=playback_ready_payload,
+        sample_rate=normalized_sample_rate,
+        channels=normalized_channels,
+    )
+
+
+__all__ = [
+    "DEFAULT_WAKE_WORD_SOUND_CONTENT_TYPE",
+    "DEFAULT_WAKE_WORD_SOUND_FILE_ID",
+    "WAKEWORD_SOUND_PATH_ENV_VAR",
+    "WAKEWORD_SOUND_MIN_DURATION_MS",
+    "WAKEWORD_SOUND_POSTROLL_MS",
+    "WAKEWORD_SOUND_PREROLL_MS",
+    "WakeWordSound",
+    "load_wake_word_detected_sound_from_env",
+]
diff --git a/stackchan_server/ws_proxy.py b/stackchan_server/ws_proxy.py
index 1c45236..39dcbab 100644
--- a/stackchan_server/ws_proxy.py
+++ b/stackchan_server/ws_proxy.py
@@ -21,11 +21,18 @@
     encode_server_metadata_message,
     encode_servo_command_message,
     encode_state_command_message,
+    encode_stored_file_data_message,
+    encode_stored_file_end_message,
+    encode_stored_file_start_message,
     parse_websocket_message,
 )
 from .speak import SpeakHandler
 from .static import LISTEN_AUDIO_FORMAT
 from .types import SpeechRecognizer, SpeechSynthesizer
+from .wakeword_sound import (
+    WAKEWORD_SOUND_PATH_ENV_VAR,
+    load_wake_word_detected_sound_from_env,
+)
 
 logger = getLogger(__name__)
 
@@ -35,6 +42,7 @@
 _RECORDINGS_DIR = _BASE_DIR / "recordings"
 
 _DOWN_WAV_CHUNK = 4096  # bytes per WebSocket frame for synthesized audio (raw PCM)
+_DOWN_FILE_CHUNK = 4096  # bytes per WebSocket frame for stored-file transfer
 _DOWN_SEGMENT_MILLIS = (
     2000  # duration of a single START-DATA-END segment in milliseconds
 )
@@ -191,6 +199,81 @@ async def speak(self, text: str) -> None:
     async def send_state_command(self, state_id: int | FirmwareState) -> None:
         await self._send_state_command(state_id)
 
+    async def send_file(
+        self,
+        *,
+        file_id: str,
+        content_type: str,
+        payload: bytes,
+        sample_rate: int = 0,
+        channels: int = 0,
+    ) -> None:
+        if not file_id:
+            raise ValueError("file_id must not be empty")
+        if not content_type:
+            raise ValueError("content_type must not be empty")
+
+        logger.info(
+            "Sending stored file id=%s type=%s bytes=%d sample_rate=%d channels=%d",
+            file_id,
+            content_type,
+            len(payload),
+            sample_rate,
+            channels,
+        )
+        await self.ws.send_bytes(
+            encode_stored_file_start_message(
+                self._next_down_seq(),
+                file_id=file_id,
+                content_type=content_type,
+                total_size=len(payload),
+                sample_rate=sample_rate,
+                channels=channels,
+            )
+        )
+        offset = 0
+        chunk_count = 0
+        while offset < len(payload):
+            chunk = payload[offset : offset + _DOWN_FILE_CHUNK]
+            await self.ws.send_bytes(
+                encode_stored_file_data_message(self._next_down_seq(), chunk)
+            )
+            offset += len(chunk)
+            chunk_count += 1
+        logger.info(
+            "Stored file payload sent id=%s chunks=%d bytes=%d",
+            file_id,
+            chunk_count,
+            len(payload),
+        )
+        await self.ws.send_bytes(encode_stored_file_end_message(self._next_down_seq()))
+        logger.info("Stored file transfer completed id=%s", file_id)
+
+    async def send_default_wake_word_sound(self) -> None:
+        sound = load_wake_word_detected_sound_from_env()
+        if sound is None:
+            logger.info(
+                "Wake-word sound upload skipped because %s is not set",
+                WAKEWORD_SOUND_PATH_ENV_VAR,
+            )
+            return
+
+        logger.info(
+            "Uploading wake-word sound id=%s sample_rate=%d channels=%d bytes=%d",
+            sound.file_id,
+            sound.sample_rate,
+            sound.channels,
+            len(sound.payload),
+        )
+
+        await self.send_file(
+            file_id=sound.file_id,
+            content_type=sound.content_type,
+            payload=sound.payload,
+            sample_rate=sound.sample_rate,
+            channels=sound.channels,
+        )
+
     async def reset_state(self) -> None:
         await self.send_state_command(FirmwareState.IDLE)