Sync upstream: 四次元壁鉴权模式、修复世界推演

2026-02-21 20:50:53 +08:00
parent adcfc38ded 4ee528621f
commit 41fa00c2d0
15 changed files with 895 additions and 809 deletions
--- a/core/iframe-messaging.js
+++ b/core/iframe-messaging.js
@@ -2,9 +2,7 @@ export function getTrustedOrigin() {
    return window.location.origin;
 }

-export function getIframeTargetOrigin(iframe) {
-    const sandbox = iframe?.getAttribute?.('sandbox') || '';
-    if (sandbox && !sandbox.includes('allow-same-origin')) return 'null';
+export function getIframeTargetOrigin() {
    return getTrustedOrigin();
 }

--- a/index.js
+++ b/index.js
@@ -31,7 +31,6 @@ import { initEnaPlanner } from "./modules/ena-planner/ena-planner.js";

 extension_settings[EXT_ID] = extension_settings[EXT_ID] || {
    enabled: true,
-    sandboxMode: false,
    recorded: { enabled: true },
    templateEditor: { enabled: true, characterBindings: {} },
    tasks: { enabled: true, globalTasks: [], processedMessages: [], character_allowed_tasks: [] },
@@ -271,7 +270,7 @@ async function waitForElement(selector, root = document, timeout = 10000) {

 function toggleSettingsControls(enabled) {
    const controls = [
-        'xiaobaix_sandbox', 'xiaobaix_recorded_enabled', 'xiaobaix_preview_enabled',
+        'xiaobaix_recorded_enabled', 'xiaobaix_preview_enabled',
        'scheduled_tasks_enabled', 'xiaobaix_template_enabled',
        'xiaobaix_immersive_enabled', 'xiaobaix_fourth_wall_enabled',
        'xiaobaix_audio_enabled', 'xiaobaix_variables_panel_enabled',
@@ -380,12 +379,6 @@ async function setupSettings() {

        if (!settings.enabled) toggleSettingsControls(false);

-        $("#xiaobaix_sandbox").prop("checked", settings.sandboxMode).on("change", async function () {
-            if (!isXiaobaixEnabled) return;
-            settings.sandboxMode = $(this).prop("checked");
-            saveSettingsDebounced();
-        });
-
        const moduleConfigs = [
            { id: 'xiaobaix_recorded_enabled', key: 'recorded' },
            { id: 'xiaobaix_immersive_enabled', key: 'immersive', init: initImmersiveMode },
@@ -532,7 +525,6 @@ async function setupSettings() {
            }
            ON.forEach(k => setChecked(MAP[k], true));
            OFF.forEach(k => setChecked(MAP[k], false));
-            setChecked('xiaobaix_sandbox', false);
            setChecked('xiaobaix_use_blob', false);
            setChecked('Wrapperiframe', true);
            try { saveSettingsDebounced(); } catch (e) {}
--- a/modules/fourth-wall/fourth-wall.html
+++ b/modules/fourth-wall/fourth-wall.html
@@ -480,16 +480,7 @@ html, body {
            <div class="fw-settings-row">
                <div class="fw-field">
                    <input type="checkbox" id="voice-enabled">
-                    <label for="voice-enabled">允许语音</label>
-                </div>
-                <div class="fw-field fw-voice-select-wrap" style="display: none;">
-                    <label>声音</label>
-                    <select id="voice-select"></select>
-                </div>
-                <div class="fw-field fw-voice-speed-wrap" style="display: none;">
-                    <label>语速</label>
-                    <input type="range" id="voice-speed" min="0.5" max="2.0" step="0.1" value="1.0" style="width:70px;">
-                    <span class="speed-val" id="voice-speed-val">1.0x</span>
+                    <label for="voice-enabled">允许语音（使用 TTS 模块音色）</label>
                </div>
            </div>

@@ -561,17 +552,6 @@ html, body {
   配置
   ══════════════════════════════════════════════════════════════════════════════ */

-const TTS_WORKER_URL = 'https://hstts.velure.codes';
-
-const VALID_EMOTIONS = ['happy', 'sad', 'angry', 'surprise', 'scare', 'hate'];
-const EMOTION_ICONS = {
-    happy: '😄', sad: '😢', angry: '😠', surprise: '😮', scare: '😨', hate: '🤢'
-};
-
-// 动态加载的声音列表
-let voiceList = [];
-let defaultVoiceKey = 'female_1';
-
 /* ══════════════════════════════════════════════════════════════════════════════
   工具函数
   ══════════════════════════════════════════════════════════════════════════════ */
@@ -618,8 +598,8 @@ function postToParent(payload) {
    window.parent.postMessage({ source: 'LittleWhiteBox-FourthWall', ...payload }, PARENT_ORIGIN);
 }

-function getEmotionIcon(emotion) {
-    return EMOTION_ICONS[emotion] || '';
+function getEmotionIcon() {
+    return '';
 }

 /* ══════════════════════════════════════════════════════════════════════════════
@@ -636,30 +616,19 @@ let state = {
    sessions: [],
    activeSessionId: null,
    imgSettings: { enablePrompt: false },
-    voiceSettings: { enabled: false, voice: 'female_1', speed: 1.0 },
+    voiceSettings: { enabled: false },
    commentarySettings: { enabled: false, probability: 30 },
    promptTemplates: {}
 };

-let currentAudio = null;
+let activeVoiceRequestId = null;

 /* ══════════════════════════════════════════════════════════════════════════════
   加载声音列表
   ══════════════════════════════════════════════════════════════════════════════ */

-async function loadVoices() {
-    try {
-        const res = await fetch(`${TTS_WORKER_URL}/voices`);
-        if (!res.ok) throw new Error('Failed to load voices');
-        const data = await res.json();
-        voiceList = data.voices || [];
-        defaultVoiceKey = data.defaultVoice || 'female_1';
-        renderVoiceSelect();
-    } catch (err) {
-        console.error('[FW Voice] 加载声音列表失败:', err);
-        // 降级：使用空列表
-        voiceList = [];
-    }
+function generateVoiceRequestId() {
+    return 'fwv_' + Date.now() + '_' + Math.random().toString(36).slice(2, 8);
 }

 /* ══════════════════════════════════════════════════════════════════════════════
@@ -786,54 +755,48 @@ function bindRetryButton(slot) {
   语音处理
   ══════════════════════════════════════════════════════════════════════════════ */

-async function playVoice(text, emotion, bubbleEl) {
-    if (currentAudio) {
-        currentAudio.pause();
-        currentAudio = null;
-        document.querySelectorAll('.fw-voice-bubble.playing').forEach(el => el.classList.remove('playing'));
-    }
+function requestPlayVoice(text, emotion, bubbleEl) {
+    // Clear previous bubble state before issuing a new request.
+    document.querySelectorAll('.fw-voice-bubble.playing, .fw-voice-bubble.loading').forEach(el => {
+        el.classList.remove('playing', 'loading');
+    });

+    const voiceRequestId = generateVoiceRequestId();
+    activeVoiceRequestId = voiceRequestId;
+    bubbleEl.dataset.voiceRequestId = voiceRequestId;
    bubbleEl.classList.add('loading');
    bubbleEl.classList.remove('error');

-    try {
+    postToParent({ type: 'PLAY_VOICE', text, emotion, voiceRequestId });
+}

-        const requestBody = {
-            voiceKey: state.voiceSettings.voice || defaultVoiceKey,
-            text: text,
-            speed: state.voiceSettings.speed || 1.0,
-            uid: 'fw_' + Date.now(),
-            reqid: generateUUID()
-        };
+function handleVoiceState(data) {
+    const { voiceRequestId, state: voiceState, duration } = data;
+    const bubble = document.querySelector(`.fw-voice-bubble[data-voice-request-id="${voiceRequestId}"]`);
+    if (!bubble) return;

-        if (emotion && VALID_EMOTIONS.includes(emotion)) {
-            requestBody.emotion = emotion;
-            requestBody.emotionScale = 5;
+    switch (voiceState) {
+        case 'loading':
+            bubble.classList.add('loading');
+            bubble.classList.remove('playing', 'error');
+            break;
+        case 'playing':
+            bubble.classList.remove('loading', 'error');
+            bubble.classList.add('playing');
+            if (duration != null) {
+                const durationEl = bubble.querySelector('.fw-voice-duration');
+                if (durationEl) durationEl.textContent = Math.ceil(duration) + '"';
            }
-
-        const res = await fetch(TTS_WORKER_URL, {
-            method: 'POST',
-            headers: { 'Content-Type': 'application/json' },
-            body: JSON.stringify(requestBody)
-        });
-
-        if (!res.ok) throw new Error(`HTTP ${res.status}`);
-
-        const data = await res.json();
-        if (data.code !== 3000) throw new Error(data.message || 'TTS失败');
-
-        bubbleEl.classList.remove('loading');
-        bubbleEl.classList.add('playing');
-
-        currentAudio = new Audio(`data:audio/mp3;base64,${data.data}`);
-        currentAudio.onended = () => { bubbleEl.classList.remove('playing'); currentAudio = null; };
-        currentAudio.onerror = () => { bubbleEl.classList.remove('playing'); bubbleEl.classList.add('error'); currentAudio = null; };
-        await currentAudio.play();
-    } catch (err) {
-        console.error('[FW Voice] TTS错误:', err);
-        bubbleEl.classList.remove('loading', 'playing');
-        bubbleEl.classList.add('error');
-        setTimeout(() => bubbleEl.classList.remove('error'), 3000);
+            break;
+        case 'ended':
+        case 'stopped':
+            bubble.classList.remove('loading', 'playing');
+            break;
+        case 'error':
+            bubble.classList.remove('loading', 'playing');
+            bubble.classList.add('error');
+            setTimeout(() => bubble.classList.remove('error'), 3000);
+            break;
    }
 }

@@ -846,12 +809,10 @@ function hydrateVoiceSlots(container) {
            bubble.onclick = e => {
                e.stopPropagation();
                if (bubble.classList.contains('loading')) return;
-                if (bubble.classList.contains('playing') && currentAudio) {
-                    currentAudio.pause();
-                    currentAudio = null;
-                    bubble.classList.remove('playing');
+                if (bubble.classList.contains('playing')) {
+                    postToParent({ type: 'STOP_VOICE', voiceRequestId: bubble.dataset.voiceRequestId });
                } else {
-                    playVoice(text, emotion, bubble);
+                    requestPlayVoice(text, emotion, bubble);
                }
            };
        }
@@ -1022,24 +983,6 @@ function renderSessionSelect() {
    ).join('');
 }

-// 使用动态加载的声音列表渲染下拉框
-function renderVoiceSelect() {
-    const select = document.getElementById('voice-select');
-    if (!select || !voiceList.length) return;
-
-    const females = voiceList.filter(v => v.gender === 'female');
-    const males = voiceList.filter(v => v.gender === 'male');
-
-    select.innerHTML = `
-        <optgroup label="👩 女声">
-            ${females.map(v => `<option value="${v.key}">${v.name} · ${v.tag}</option>`).join('')}
-        </optgroup>
-        <optgroup label="👨 男声">
-            ${males.map(v => `<option value="${v.key}">${v.name} · ${v.tag}</option>`).join('')}
-        </optgroup>
-    `;
-    select.value = state.voiceSettings.voice || defaultVoiceKey;
-}

 function updateMenuUI() {
    const actions = document.getElementById('header-actions');
@@ -1057,11 +1000,6 @@ function updateMenuUI() {
    }
 }

-function updateVoiceUI(enabled) {
-    document.querySelector('.fw-voice-select-wrap').style.display = enabled ? '' : 'none';
-    document.querySelector('.fw-voice-speed-wrap').style.display = enabled ? '' : 'none';
-}
-
 function updateCommentaryUI(enabled) {
    document.querySelector('.fw-commentary-prob-wrap').style.display = enabled ? '' : 'none';
 }
@@ -1161,14 +1099,6 @@ window.addEventListener('message', event => {
            document.getElementById('img-prompt-enabled').checked = state.imgSettings.enablePrompt;
            document.getElementById('voice-enabled').checked = state.voiceSettings.enabled;

-            // 等声音列表加载完再设置值
-            if (voiceList.length) {
-                document.getElementById('voice-select').value = state.voiceSettings.voice || defaultVoiceKey;
-            }
-            document.getElementById('voice-speed').value = state.voiceSettings.speed || 1.0;
-            document.getElementById('voice-speed-val').textContent = (state.voiceSettings.speed || 1.0).toFixed(1) + 'x';
-            updateVoiceUI(state.voiceSettings.enabled);
-
            document.getElementById('commentary-enabled').checked = state.commentarySettings.enabled;
            document.getElementById('commentary-prob').value = state.commentarySettings.probability;
            document.getElementById('commentary-prob-val').textContent = state.commentarySettings.probability + '%';
@@ -1211,6 +1141,10 @@ window.addEventListener('message', event => {
            updateFullscreenButton(data.isFullscreen);
            break;

+        case 'VOICE_STATE':
+            handleVoiceState(data);
+            break;
+
        case 'IMAGE_RESULT':
            handleImageResult(data);
            break;
@@ -1229,9 +1163,6 @@ window.addEventListener('message', event => {
   ══════════════════════════════════════════════════════════════════════════════ */

 document.addEventListener('DOMContentLoaded', async () => {
-    // 先加载声音列表
-    await loadVoices();
-
    document.getElementById('btn-menu-toggle').onclick = () => { state.menuExpanded = !state.menuExpanded; updateMenuUI(); };
    document.getElementById('btn-settings').onclick = () => { state.settingsOpen = !state.settingsOpen; document.getElementById('settings-panel').classList.toggle('open', state.settingsOpen); };
    document.getElementById('btn-settings-close').onclick = () => { state.settingsOpen = false; document.getElementById('settings-panel').classList.remove('open'); };
@@ -1258,19 +1189,6 @@ document.addEventListener('DOMContentLoaded', async () => {

    document.getElementById('voice-enabled').onchange = function() {
        state.voiceSettings.enabled = this.checked;
-        updateVoiceUI(this.checked);
-        postToParent({ type: 'SAVE_VOICE_SETTINGS', voiceSettings: state.voiceSettings });
-    };
-
-    document.getElementById('voice-select').onchange = function() {
-        state.voiceSettings.voice = this.value;
-        postToParent({ type: 'SAVE_VOICE_SETTINGS', voiceSettings: state.voiceSettings });
-    };
-
-    document.getElementById('voice-speed').oninput = function() {
-        const val = parseFloat(this.value);
-        document.getElementById('voice-speed-val').textContent = val.toFixed(1) + 'x';
-        state.voiceSettings.speed = val;
        postToParent({ type: 'SAVE_VOICE_SETTINGS', voiceSettings: state.voiceSettings });
    };

--- a/modules/fourth-wall/fourth-wall.js
+++ b/modules/fourth-wall/fourth-wall.js
@@ -1,5 +1,5 @@
-// ════════════════════════════════════════════════════════════════════════════
-// 次元壁模块 - 主控制器
+// ════════════════════════════════════════════
+// Fourth Wall Module - Main Controller
 // ════════════════════════════════════════════════════════════════════════════
 import { extension_settings, getContext, saveMetadataDebounced } from "../../../../../extensions.js";
 import { saveSettingsDebounced, chat_metadata, default_user_avatar, default_avatar } from "../../../../../../script.js";
@@ -8,7 +8,7 @@ import { createModuleEvents, event_types } from "../../core/event-manager.js";
 import { xbLog } from "../../core/debug-core.js";

 import { handleCheckCache, handleGenerate, clearExpiredCache } from "./fw-image.js";
-import { DEFAULT_VOICE, DEFAULT_SPEED } from "./fw-voice.js";
+import { synthesizeAndPlay, stopCurrent as stopCurrentVoice } from "./fw-voice-runtime.js";
 import {
    buildPrompt,
    buildCommentaryPrompt,
@@ -19,8 +19,9 @@ import {
 } from "./fw-prompt.js";
 import { initMessageEnhancer, cleanupMessageEnhancer } from "./fw-message-enhancer.js";
 import { postToIframe, isTrustedMessage, getTrustedOrigin } from "../../core/iframe-messaging.js";
-// ════════════════════════════════════════════════════════════════════════════
-// 常量
+
+// ════════════════════════════════════════════
+// Constants
 // ════════════════════════════════════════════════════════════════════════════

 const events = createModuleEvents('fourthWall');
@@ -30,7 +31,7 @@ const COMMENTARY_COOLDOWN = 180000;
 const IFRAME_PING_TIMEOUT = 800;

 // ════════════════════════════════════════════════════════════════════════════
-// 状态
+// State
 // ════════════════════════════════════════════════════════════════════════════

 let overlayCreated = false;
@@ -44,14 +45,13 @@ let currentLoadedChatId = null;
 let lastCommentaryTime = 0;
 let commentaryBubbleEl = null;
 let commentaryBubbleTimer = null;
+let currentVoiceRequestId = null;

-// ═══════════════════════════════ 新增 ═══════════════════════════════
 let visibilityHandler = null;
 let pendingPingId = null;
-// ════════════════════════════════════════════════════════════════════

 // ════════════════════════════════════════════════════════════════════════════
-// 设置管理（保持不变）
+// Settings
 // ════════════════════════════════════════════════════════════════════════════

 function getSettings() {
@@ -60,7 +60,7 @@ function getSettings() {

    s.fourthWall ||= { enabled: true };
    s.fourthWallImage ||= { enablePrompt: false };
-    s.fourthWallVoice ||= { enabled: false, voice: DEFAULT_VOICE, speed: DEFAULT_SPEED };
+    s.fourthWallVoice ||= { enabled: false };
    s.fourthWallCommentary ||= { enabled: false, probability: 30 };
    s.fourthWallPromptTemplates ||= {};

@@ -74,7 +74,7 @@ function getSettings() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 工具函数（保持不变）
+// Utilities
 // ════════════════════════════════════════════════════════════════════════════

 function b64UrlEncode(str) {
@@ -162,7 +162,7 @@ function getAvatarUrls() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 存储管理（保持不变）
+// Storage
 // ════════════════════════════════════════════════════════════════════════════

 function getFWStore(chatId = getCurrentChatIdSafe()) {
@@ -177,7 +177,7 @@ function getFWStore(chatId = getCurrentChatIdSafe()) {

    if (!fw.sessions) {
        const oldHistory = Array.isArray(fw.history) ? fw.history.slice() : [];
-        fw.sessions = [{ id: 'default', name: '默认记录', createdAt: Date.now(), history: oldHistory }];
+        fw.sessions = [{ id: 'default', name: 'Default', createdAt: Date.now(), history: oldHistory }];
        fw.activeSessionId = 'default';
        if (Object.prototype.hasOwnProperty.call(fw, 'history')) delete fw.history;
    }
@@ -199,7 +199,7 @@ function saveFWStore() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// iframe 通讯
+// iframe Communication
 // ════════════════════════════════════════════════════════════════════════════

 function postToFrame(payload) {
@@ -240,15 +240,13 @@ function sendInitData() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// iframe 健康检测与恢复（新增）
+// iframe Health Check & Recovery
 // ════════════════════════════════════════════════════════════════════════════

 function handleVisibilityChange() {
    if (document.visibilityState !== 'visible') return;
-    
    const overlay = document.getElementById('xiaobaix-fourth-wall-overlay');
    if (!overlay || overlay.style.display === 'none') return;
-    
    checkIframeHealth();
 }

@@ -256,35 +254,31 @@ function checkIframeHealth() {
    const iframe = document.getElementById('xiaobaix-fourth-wall-iframe');
    if (!iframe) return;

-    // 生成唯一 ping ID
    const pingId = 'ping_' + Date.now();
    pendingPingId = pingId;

-    // 尝试发送 PING
    try {
        const win = iframe.contentWindow;
        if (!win) {
-            recoverIframe('contentWindow 不存在');
+            recoverIframe('contentWindow missing');
            return;
        }
        win.postMessage({ source: 'LittleWhiteBox', type: 'PING', pingId }, getTrustedOrigin());
    } catch (e) {
-        recoverIframe('无法访问 iframe: ' + e.message);
+        recoverIframe('Cannot access iframe: ' + e.message);
        return;
    }

-    // 设置超时检测
    setTimeout(() => {
        if (pendingPingId === pingId) {
-            // 没有收到 PONG 响应
-            recoverIframe('PING 超时无响应');
+            recoverIframe('PING timeout');
        }
    }, IFRAME_PING_TIMEOUT);
 }

 function handlePongResponse(pingId) {
    if (pendingPingId === pingId) {
-        pendingPingId = null; // 清除，表示收到响应
+        pendingPingId = null;
    }
 }

@@ -292,24 +286,72 @@ function recoverIframe(reason) {
    const iframe = document.getElementById('xiaobaix-fourth-wall-iframe');
    if (!iframe) return;

-    try { xbLog.warn('fourthWall', `iframe 恢复中: ${reason}`); } catch {}
+    try { xbLog.warn('fourthWall', `iframe recovery: ${reason}`); } catch { }

-    // 重置状态
    frameReady = false;
    pendingFrameMessages = [];
    pendingPingId = null;

-    // 如果正在流式生成，取消
    if (isStreaming) {
        cancelGeneration();
    }

-    // 重新加载 iframe
    iframe.src = iframePath;
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 消息处理（添加 PONG 处理）
+// Voice Handling
+// ════════════════════════════════════════════════════════════════════════════
+
+function handlePlayVoice(data) {
+    const { text, emotion, voiceRequestId } = data;
+
+    if (!text?.trim()) {
+        postToFrame({ type: 'VOICE_STATE', voiceRequestId, state: 'error', message: 'Voice text is empty' });
+        return;
+    }
+
+    // Notify old request as stopped
+    if (currentVoiceRequestId && currentVoiceRequestId !== voiceRequestId) {
+        postToFrame({ type: 'VOICE_STATE', voiceRequestId: currentVoiceRequestId, state: 'stopped' });
+    }
+
+    currentVoiceRequestId = voiceRequestId;
+
+    synthesizeAndPlay(text, emotion, {
+        requestId: voiceRequestId,
+        onState(state, info) {
+            if (currentVoiceRequestId !== voiceRequestId) return;
+            postToFrame({
+                type: 'VOICE_STATE',
+                voiceRequestId,
+                state,
+                duration: info?.duration,
+                message: info?.message,
+            });
+        },
+    });
+}
+
+function handleStopVoice(data) {
+    const targetId = data?.voiceRequestId || currentVoiceRequestId;
+    stopCurrentVoice();
+    if (targetId) {
+        postToFrame({ type: 'VOICE_STATE', voiceRequestId: targetId, state: 'stopped' });
+    }
+    currentVoiceRequestId = null;
+}
+
+function stopVoiceAndNotify() {
+    if (currentVoiceRequestId) {
+        postToFrame({ type: 'VOICE_STATE', voiceRequestId: currentVoiceRequestId, state: 'stopped' });
+    }
+    stopCurrentVoice();
+    currentVoiceRequestId = null;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// Frame Message Handler
 // ════════════════════════════════════════════════════════════════════════════

 function handleFrameMessage(event) {
@@ -327,11 +369,9 @@ function handleFrameMessage(event) {
            sendInitData();
            break;

-        // ═══════════════════════════ 新增 ═══════════════════════════
        case 'PONG':
            handlePongResponse(data.pingId);
            break;
-        // ════════════════════════════════════════════════════════════

        case 'TOGGLE_FULLSCREEN':
            toggleFullscreen();
@@ -362,7 +402,7 @@ function handleFrameMessage(event) {
            break;

        case 'SAVE_VOICE_SETTINGS':
-            Object.assign(settings.fourthWallVoice, data.voiceSettings);
+            settings.fourthWallVoice.enabled = !!data.voiceSettings?.enabled;
            saveSettingsDebounced();
            break;

@@ -446,11 +486,19 @@ function handleFrameMessage(event) {
        case 'GENERATE_IMAGE':
            handleGenerate(data, postToFrame);
            break;
+
+        case 'PLAY_VOICE':
+            handlePlayVoice(data);
+            break;
+
+        case 'STOP_VOICE':
+            handleStopVoice(data);
+            break;
    }
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 生成处理（保持不变）
+// Generation
 // ════════════════════════════════════════════════════════════════════════════

 async function startGeneration(data) {
@@ -464,7 +512,7 @@ async function startGeneration(data) {
    });

    const gen = window.xiaobaixStreamingGeneration;
-    if (!gen?.xbgenrawCommand) throw new Error('xbgenraw 模块不可用');
+    if (!gen?.xbgenrawCommand) throw new Error('xbgenraw module unavailable');

    const topMessages = [
        { role: 'user', content: msg1 },
@@ -560,8 +608,8 @@ function stopStreamingPoll() {
 function finalizeGeneration() {
    stopStreamingPoll();
    const gen = window.xiaobaixStreamingGeneration;
-    const rawText = gen?.getLastGeneration?.(STREAM_SESSION_ID) || '(无响应)';
-    const finalText = extractMsg(rawText) || '(无响应)';
+    const rawText = gen?.getLastGeneration?.(STREAM_SESSION_ID) || '(no response)';
+    const finalText = extractMsg(rawText) || '(no response)';
    const thinkingText = extractThinking(rawText);

    isStreaming = false;
@@ -579,12 +627,12 @@ function cancelGeneration() {
    const gen = window.xiaobaixStreamingGeneration;
    stopStreamingPoll();
    isStreaming = false;
-    try { gen?.cancel?.(STREAM_SESSION_ID); } catch {}
+    try { gen?.cancel?.(STREAM_SESSION_ID); } catch { }
    postToFrame({ type: 'GENERATION_CANCELLED' });
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 实时吐槽（保持不变，省略...）
+// Commentary
 // ════════════════════════════════════════════════════════════════════════════

 function shouldTriggerCommentary() {
@@ -669,7 +717,7 @@ async function handleAIMessageForCommentary(data) {
    if (!commentary) return;
    const session = getActiveSession();
    if (session) {
-        session.history.push({ role: 'ai', content: `（瞄了眼刚才的台词）${commentary}`, ts: Date.now(), type: 'commentary' });
+        session.history.push({ role: 'ai', content: `(glanced at the last line) ${commentary}`, ts: Date.now(), type: 'commentary' });
        saveFWStore();
    }
    showCommentaryBubble(commentary);
@@ -693,7 +741,7 @@ async function handleEditForCommentary(data) {

    const session = getActiveSession();
    if (session) {
-        const prefix = editType === 'edit_ai' ? '（发现你改了我的台词）' : '（发现你偷偷改台词）';
+        const prefix = editType === 'edit_ai' ? '(noticed you edited my line) ' : '(caught you sneaking edits) ';
        session.history.push({ role: 'ai', content: `${prefix}${commentary}`, ts: Date.now(), type: 'commentary' });
        saveFWStore();
    }
@@ -705,7 +753,7 @@ function getFloatBtnPosition() {
    if (!btn) return null;
    const rect = btn.getBoundingClientRect();
    let stored = {};
-    try { stored = JSON.parse(localStorage.getItem(`${EXT_ID}:fourthWallFloatBtnPos`) || '{}') || {}; } catch {}
+    try { stored = JSON.parse(localStorage.getItem(`${EXT_ID}:fourthWallFloatBtnPos`) || '{}') || {}; } catch { }
    return { top: rect.top, left: rect.left, width: rect.width, height: rect.height, side: stored.side || 'right' };
 }

@@ -771,8 +819,8 @@ function cleanupCommentary() {
    lastCommentaryTime = 0;
 }

-// ════════════════════════════════════════════════════════════════════════════
-// Overlay 管理（添加可见性监听）
+// ════════════════════════════════════════════
+// Overlay
 // ════════════════════════════════════════════════════════════════════════════

 function createOverlay() {
@@ -822,25 +870,22 @@ function showOverlay() {
    sendInitData();
    postToFrame({ type: 'FULLSCREEN_STATE', isFullscreen: !!document.fullscreenElement });

-    // ═══════════════════════════ 新增：添加可见性监听 ═══════════════════════════
    if (!visibilityHandler) {
        visibilityHandler = handleVisibilityChange;
        document.addEventListener('visibilitychange', visibilityHandler);
    }
-    // ════════════════════════════════════════════════════════════════════════════
 }

 function hideOverlay() {
    $('#xiaobaix-fourth-wall-overlay').hide();
-    if (document.fullscreenElement) document.exitFullscreen().catch(() => {});
+    if (document.fullscreenElement) document.exitFullscreen().catch(() => { });
+    stopVoiceAndNotify();

-    // ═══════════════════════════ 新增：移除可见性监听 ═══════════════════════════
    if (visibilityHandler) {
        document.removeEventListener('visibilitychange', visibilityHandler);
        visibilityHandler = null;
    }
    pendingPingId = null;
-    // ════════════════════════════════════════════════════════════════════════════
 }

 function toggleFullscreen() {
@@ -850,16 +895,16 @@ function toggleFullscreen() {
    if (document.fullscreenElement) {
        document.exitFullscreen().then(() => {
            postToFrame({ type: 'FULLSCREEN_STATE', isFullscreen: false });
-        }).catch(() => {});
+        }).catch(() => { });
    } else if (overlay.requestFullscreen) {
        overlay.requestFullscreen().then(() => {
            postToFrame({ type: 'FULLSCREEN_STATE', isFullscreen: true });
-        }).catch(() => {});
+        }).catch(() => { });
    }
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 悬浮按钮（保持不变，省略...）
+// Floating Button
 // ════════════════════════════════════════════════════════════════════════════

 function createFloatingButton() {
@@ -871,7 +916,7 @@ function createFloatingButton() {

    const clamp = (v, min, max) => Math.min(Math.max(v, min), max);
    const readPos = () => { try { return JSON.parse(localStorage.getItem(POS_KEY) || 'null'); } catch { return null; } };
-    const writePos = (pos) => { try { localStorage.setItem(POS_KEY, JSON.stringify(pos)); } catch {} };
+    const writePos = (pos) => { try { localStorage.setItem(POS_KEY, JSON.stringify(pos)); } catch { } };
    const calcDockLeft = (side, w) => (side === 'left' ? -Math.round(w / 2) : (window.innerWidth - Math.round(w / 2)));
    const applyDocked = (side, topRatio) => {
        const btn = document.getElementById('xiaobaix-fw-float-btn');
@@ -885,7 +930,7 @@ function createFloatingButton() {
    };

    const $btn = $(`
-        <button id="xiaobaix-fw-float-btn" title="皮下交流" style="position:fixed!important;left:0px!important;top:0px!important;z-index:9999!important;width:${size}px!important;height:${size}px!important;border-radius:50%!important;border:none!important;background:linear-gradient(135deg,#667eea 0%,#764ba2 100%)!important;color:#fff!important;font-size:${Math.round(size * 0.45)}px!important;cursor:pointer!important;box-shadow:0 4px 15px rgba(102,126,234,0.4)!important;display:flex!important;align-items:center!important;justify-content:center!important;transition:left 0.2s,top 0.2s,transform 0.2s,box-shadow 0.2s!important;touch-action:none!important;user-select:none!important;">
+        <button id="xiaobaix-fw-float-btn" title="Fourth Wall" style="position:fixed!important;left:0px!important;top:0px!important;z-index:9999!important;width:${size}px!important;height:${size}px!important;border-radius:50%!important;border:none!important;background:linear-gradient(135deg,#667eea 0%,#764ba2 100%)!important;color:#fff!important;font-size:${Math.round(size * 0.45)}px!important;cursor:pointer!important;box-shadow:0 4px 15px rgba(102,126,234,0.4)!important;display:flex!important;align-items:center!important;justify-content:center!important;transition:left 0.2s,top 0.2s,transform 0.2s,box-shadow 0.2s!important;touch-action:none!important;user-select:none!important;">
            <i class="fa-solid fa-comments"></i>
        </button>
    `);
@@ -896,8 +941,8 @@ function createFloatingButton() {
        showOverlay();
    });

-    $btn.on('mouseenter', function() { $(this).css({ 'transform': 'scale(1.08)', 'box-shadow': '0 6px 20px rgba(102, 126, 234, 0.5)' }); });
-    $btn.on('mouseleave', function() { $(this).css({ 'transform': 'none', 'box-shadow': '0 4px 15px rgba(102, 126, 234, 0.4)' }); });
+    $btn.on('mouseenter', function () { $(this).css({ 'transform': 'scale(1.08)', 'box-shadow': '0 6px 20px rgba(102, 126, 234, 0.5)' }); });
+    $btn.on('mouseleave', function () { $(this).css({ 'transform': 'none', 'box-shadow': '0 4px 15px rgba(102, 126, 234, 0.4)' }); });

    document.body.appendChild($btn[0]);

@@ -911,7 +956,7 @@ function createFloatingButton() {
        if (e.button !== undefined && e.button !== 0) return;
        const btn = e.currentTarget;
        pointerId = e.pointerId;
-        try { btn.setPointerCapture(pointerId); } catch {}
+        try { btn.setPointerCapture(pointerId); } catch { }
        const rect = btn.getBoundingClientRect();
        startX = e.clientX; startY = e.clientY; startLeft = rect.left; startTop = rect.top;
        dragging = false;
@@ -937,7 +982,7 @@ function createFloatingButton() {
    const onPointerUp = (e) => {
        if (pointerId === null || e.pointerId !== pointerId) return;
        const btn = e.currentTarget;
-        try { btn.releasePointerCapture(pointerId); } catch {}
+        try { btn.releasePointerCapture(pointerId); } catch { }
        pointerId = null;
        btn.style.transition = '';
        const rect = btn.getBoundingClientRect();
@@ -976,12 +1021,12 @@ function removeFloatingButton() {
    }
 }

-// ════════════════════════════════════════════════════════════════════════════
-// 初始化和清理
-// ════════════════════════════════════════════════════════════════════════════
+// ════════════════════════════════════════════
+// Init & Cleanup
+// ════════════════════════════════════════════

 function initFourthWall() {
-    try { xbLog.info('fourthWall', 'initFourthWall'); } catch {}
+    try { xbLog.info('fourthWall', 'initFourthWall'); } catch { }
    const settings = getSettings();
    if (!settings.fourthWall?.enabled) return;

@@ -999,13 +1044,15 @@ function initFourthWall() {
 }

 function fourthWallCleanup() {
-    try { xbLog.info('fourthWall', 'fourthWallCleanup'); } catch {}
+    try { xbLog.info('fourthWall', 'fourthWallCleanup'); } catch { }
    events.cleanup();
    cleanupCommentary();
    removeFloatingButton();
    hideOverlay();
    cancelGeneration();
    cleanupMessageEnhancer();
+    stopCurrentVoice();
+    currentVoiceRequestId = null;
    frameReady = false;
    pendingFrameMessages = [];
    overlayCreated = false;
@@ -1029,7 +1076,7 @@ if (typeof window !== 'undefined') {

    document.addEventListener('xiaobaixEnabledChanged', e => {
        if (e?.detail?.enabled === false) {
-            try { fourthWallCleanup(); } catch {}
+            try { fourthWallCleanup(); } catch { }
        }
    });
 }
--- a/modules/fourth-wall/fw-message-enhancer.js
+++ b/modules/fourth-wall/fw-message-enhancer.js
@@ -1,5 +1,5 @@
 // ════════════════════════════════════════════════════════════════════════════
-// 消息楼层增强器
+// Message Floor Enhancer
 // ════════════════════════════════════════════════════════════════════════════

 import { extension_settings } from "../../../../../extensions.js";
@@ -8,37 +8,29 @@ import { createModuleEvents, event_types } from "../../core/event-manager.js";
 import { xbLog } from "../../core/debug-core.js";

 import { generateImage, clearQueue } from "./fw-image.js";
-import { 
-    synthesizeSpeech, 
-    loadVoices,
-    VALID_EMOTIONS, 
-    DEFAULT_VOICE, 
-    DEFAULT_SPEED 
-} from "./fw-voice.js";
+import { synthesizeAndPlay, stopCurrent as stopCurrentVoice } from "./fw-voice-runtime.js";

-// ════════════════════════════════════════════════════════════════════════════
-// 状态
+// ════════════════════════════════════════════
+// State
 // ════════════════════════════════════════════════════════════════════════════

 const events = createModuleEvents('messageEnhancer');
 const CSS_INJECTED_KEY = 'xb-me-css-injected';

-let currentAudio = null;
 let imageObserver = null;
 let novelDrawObserver = null;

-// ════════════════════════════════════════════════════════════════════════════
-// 初始化与清理
-// ════════════════════════════════════════════════════════════════════════════
+// ════════════════════════════════════════════
+// Init & Cleanup
+// ════════════════════════════════════════════

 export async function initMessageEnhancer() {
    const settings = extension_settings[EXT_ID];
    if (!settings?.fourthWall?.enabled) return;

-    xbLog.info('messageEnhancer', '初始化消息增强器');
+    xbLog.info('messageEnhancer', 'init message enhancer');

    injectStyles();
-    await loadVoices();
    initImageObserver();
    initNovelDrawObserver();

@@ -60,11 +52,13 @@ export async function initMessageEnhancer() {
 }

 export function cleanupMessageEnhancer() {
-    xbLog.info('messageEnhancer', '清理消息增强器');
+    xbLog.info('messageEnhancer', 'cleanup message enhancer');

    events.cleanup();
    clearQueue();

+    stopCurrentVoice();
+
    if (imageObserver) {
        imageObserver.disconnect();
        imageObserver = null;
@@ -74,15 +68,10 @@ export function cleanupMessageEnhancer() {
        novelDrawObserver.disconnect();
        novelDrawObserver = null;
    }
-    
-    if (currentAudio) {
-        currentAudio.pause();
-        currentAudio = null;
-    }
 }

-// ════════════════════════════════════════════════════════════════════════════
-// NovelDraw 兼容
+// ════════════════════════════════════════════
+// NovelDraw Compat
 // ════════════════════════════════════════════════════════════════════════════

 function initNovelDrawObserver() {
@@ -135,8 +124,8 @@ function hasUnrenderedVoice(mesText) {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 事件处理
-// ════════════════════════════════════════════════════════════════════════════
+// Event Handlers
+// ════════════════════════════════════════════

 function handleMessageChange(data) {
    setTimeout(() => {
@@ -160,7 +149,7 @@ function processAllMessages() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 图片观察器
+// Image Observer
 // ════════════════════════════════════════════════════════════════════════════

 function initImageObserver() {
@@ -180,8 +169,8 @@ function initImageObserver() {
 }

 // ════════════════════════════════════════════════════════════════════════════
-// 样式注入
-// ════════════════════════════════════════════════════════════════════════════
+// Style Injection
+// ════════════════════════════════════════════

 function injectStyles() {
    if (document.getElementById(CSS_INJECTED_KEY)) return;
@@ -251,8 +240,8 @@ function injectStyles() {
    document.head.appendChild(style);
 }

-// ════════════════════════════════════════════════════════════════════════════
-// 内容增强
+// ════════════════════════════════════════════
+// Content Enhancement
 // ════════════════════════════════════════════════════════════════════════════

 function enhanceMessageContent(container) {
@@ -313,8 +302,8 @@ function escapeHtml(text) {
    return String(text || '').replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
 }

-// ════════════════════════════════════════════════════════════════════════════
-// 图片处理
+// ════════════════════════════════════════════
+// Image Handling
 // ════════════════════════════════════════════════════════════════════════════

 function hydrateImageSlots(container) {
@@ -323,7 +312,6 @@ function hydrateImageSlots(container) {
        slot.dataset.observed = '1';

        if (!slot.dataset.loaded && !slot.dataset.loading && !slot.querySelector('img')) {
-            // Template-only UI markup.
            // eslint-disable-next-line no-unsanitized/property
            slot.innerHTML = `<div class="xb-img-placeholder"><i class="fa-regular fa-image"></i><span>滚动加载</span></div>`;
        }
@@ -333,7 +321,6 @@ function hydrateImageSlots(container) {
 }

 async function loadImage(slot, tags) {
-    // Template-only UI markup.
    // eslint-disable-next-line no-unsanitized/property
    slot.innerHTML = `<div class="xb-img-loading"><i class="fa-solid fa-spinner"></i> 检查缓存...</div>`;

@@ -341,17 +328,14 @@ async function loadImage(slot, tags) {
        const base64 = await generateImage(tags, (status, position, delay) => {
            switch (status) {
                case 'queued':
-                    // Template-only UI markup.
                    // eslint-disable-next-line no-unsanitized/property
                    slot.innerHTML = `<div class="xb-img-loading"><i class="fa-solid fa-clock"></i> 排队中 #${position}</div>`;
                    break;
                case 'generating':
-                    // Template-only UI markup.
                    // eslint-disable-next-line no-unsanitized/property
                    slot.innerHTML = `<div class="xb-img-loading"><i class="fa-solid fa-palette"></i> 生成中${position > 0 ? ` (${position} 排队)` : ''}...</div>`;
                    break;
                case 'waiting':
-                    // Template-only UI markup.
                    // eslint-disable-next-line no-unsanitized/property
                    slot.innerHTML = `<div class="xb-img-loading"><i class="fa-solid fa-clock"></i> 排队中 #${position} (${delay}s)</div>`;
                    break;
@@ -365,7 +349,6 @@ async function loadImage(slot, tags) {
        slot.dataset.loading = '';

        if (err.message === '队列已清空') {
-            // Template-only UI markup.
            // eslint-disable-next-line no-unsanitized/property
            slot.innerHTML = `<div class="xb-img-placeholder"><i class="fa-regular fa-image"></i><span>滚动加载</span></div>`;
            slot.dataset.loading = '';
@@ -373,7 +356,6 @@ async function loadImage(slot, tags) {
            return;
        }

-        // Template-only UI markup with escaped error text.
        // eslint-disable-next-line no-unsanitized/property
        slot.innerHTML = `<div class="xb-img-error"><i class="fa-solid fa-exclamation-triangle"></i><div>${escapeHtml(err?.message || '失败')}</div><button class="xb-img-retry" data-tags="${encodeURIComponent(tags)}">重试</button></div>`;
        bindRetryButton(slot);
@@ -389,7 +371,6 @@ function renderImage(slot, base64, fromCache) {
    img.className = 'xb-generated-img';
    img.onclick = () => window.open(img.src, '_blank');

-    // Template-only UI markup.
    // eslint-disable-next-line no-unsanitized/property
    slot.innerHTML = '';
    slot.appendChild(img);
@@ -397,7 +378,6 @@ function renderImage(slot, base64, fromCache) {
    if (fromCache) {
        const badge = document.createElement('span');
        badge.className = 'xb-img-badge';
-        // Template-only UI markup.
        // eslint-disable-next-line no-unsanitized/property
        badge.innerHTML = '<i class="fa-solid fa-bolt"></i>';
        slot.appendChild(badge);
@@ -417,8 +397,8 @@ function bindRetryButton(slot) {
    };
 }

-// ════════════════════════════════════════════════════════════════════════════
-// 语音处理
+// ════════════════════════════════════════════
+// Voice Handling
 // ════════════════════════════════════════════════════════════════════════════

 function hydrateVoiceSlots(container) {
@@ -434,48 +414,43 @@ function hydrateVoiceSlots(container) {
            e.stopPropagation();
            if (bubble.classList.contains('loading')) return;

-            if (bubble.classList.contains('playing') && currentAudio) {
-                currentAudio.pause();
-                currentAudio = null;
+            if (bubble.classList.contains('playing')) {
+                stopCurrentVoice();
                bubble.classList.remove('playing');
                return;
            }

-            if (currentAudio) {
-                currentAudio.pause();
-                currentAudio = null;
-            }
-            document.querySelectorAll('.xb-voice-bubble.playing').forEach(el => el.classList.remove('playing'));
+            // Clear other bubble states
+            document.querySelectorAll('.xb-voice-bubble.playing, .xb-voice-bubble.loading').forEach(el => {
+                el.classList.remove('playing', 'loading');
+            });

-            await playVoice(text, emotion, bubble);
+            bubble.classList.add('loading');
+            bubble.classList.remove('error');
+
+            synthesizeAndPlay(text, emotion, {
+                onState(state, info) {
+                    switch (state) {
+                        case 'loading':
+                            bubble.classList.add('loading');
+                            bubble.classList.remove('playing', 'error');
+                            break;
+                        case 'playing':
+                            bubble.classList.remove('loading', 'error');
+                            bubble.classList.add('playing');
+                            break;
+                        case 'ended':
+                        case 'stopped':
+                            bubble.classList.remove('loading', 'playing');
+                            break;
+                        case 'error':
+                            bubble.classList.remove('loading', 'playing');
+                            bubble.classList.add('error');
+                            setTimeout(() => bubble.classList.remove('error'), 3000);
+                            break;
+                    }
+                },
+            });
        };
    });
 }
-
-async function playVoice(text, emotion, bubbleEl) {
-    bubbleEl.classList.add('loading');
-    bubbleEl.classList.remove('error');
-    
-    try {
-        const settings = extension_settings[EXT_ID]?.fourthWallVoice || {};
-        const audioBase64 = await synthesizeSpeech(text, {
-            voiceKey: settings.voice || DEFAULT_VOICE,
-            speed: settings.speed || DEFAULT_SPEED,
-            emotion: VALID_EMOTIONS.includes(emotion) ? emotion : null
-        });
-        
-        bubbleEl.classList.remove('loading');
-        bubbleEl.classList.add('playing');
-        
-        currentAudio = new Audio(`data:audio/mp3;base64,${audioBase64}`);
-        currentAudio.onended = () => { bubbleEl.classList.remove('playing'); currentAudio = null; };
-        currentAudio.onerror = () => { bubbleEl.classList.remove('playing'); bubbleEl.classList.add('error'); currentAudio = null; };
-        await currentAudio.play();
-        
-    } catch (err) {
-        console.error('[MessageEnhancer] TTS 错误:', err);
-        bubbleEl.classList.remove('loading', 'playing');
-        bubbleEl.classList.add('error');
-        setTimeout(() => bubbleEl.classList.remove('error'), 3000);
-    }
-}
--- a/modules/fourth-wall/fw-voice-runtime.js
+++ b/modules/fourth-wall/fw-voice-runtime.js
@@ -0,0 +1,132 @@
+// ════════════════════════════════════════════
+// 语音运行时 - 统一合成与互斥播放
+// ════════════════════════════════════════════
+
+import { normalizeEmotion } from '../tts/tts-text.js';
+
+let currentAudio = null;
+let currentObjectUrl = null;
+let currentAbort = null;
+let currentRequestId = null;
+
+/**
+ * 合成并播放语音（后播覆盖前播）
+ *
+ * @param {string} text - 要合成的文本
+ * @param {string} [emotion] - 原始情绪字符串（自动 normalize）
+ * @param {Object} [callbacks] - 状态回调
+ * @param {string} [callbacks.requestId] - 请求标识，用于防时序错乱
+ * @param {(state: string, info?: object) => void} [callbacks.onState] - 状态变化回调
+ *   state: 'loading' | 'playing' | 'ended' | 'error' | 'stopped'
+ *   info: { duration?, message? }
+ * @returns {{ stop: () => void }} 控制句柄
+ */
+export function synthesizeAndPlay(text, emotion, callbacks) {
+    const requestId = callbacks?.requestId || `vr_${Date.now()}_${Math.random().toString(36).slice(2, 8)}`;
+    const onState = callbacks?.onState;
+
+    // 停掉前一个（回传 stopped）
+    stopCurrent();
+
+    const abortController = new AbortController();
+    currentAbort = abortController;
+    currentRequestId = requestId;
+
+    const notify = (state, info) => {
+        if (currentRequestId !== requestId && state !== 'stopped') return;
+        onState?.(state, info);
+    };
+
+    notify('loading');
+
+    const run = async () => {
+        const synthesize = window.xiaobaixTts?.synthesize;
+        if (typeof synthesize !== 'function') {
+            throw new Error('请先启用 TTS 模块');
+        }
+
+        const blob = await synthesize(text, {
+            emotion: normalizeEmotion(emotion || ''),
+            signal: abortController.signal,
+        });
+
+        if (abortController.signal.aborted) return;
+        if (currentRequestId !== requestId) return;
+
+        const url = URL.createObjectURL(blob);
+        const audio = new Audio(url);
+
+        // 清理旧的（理论上 stopCurrent 已清理，防御性）
+        cleanup();
+
+        currentAudio = audio;
+        currentObjectUrl = url;
+
+        audio.onloadedmetadata = () => {
+            if (currentRequestId !== requestId) return;
+            notify('playing', { duration: audio.duration || 0 });
+        };
+
+        audio.onended = () => {
+            if (currentRequestId !== requestId) return;
+            notify('ended');
+            cleanup();
+        };
+
+        audio.onerror = () => {
+            if (currentRequestId !== requestId) return;
+            notify('error', { message: '播放失败' });
+            cleanup();
+        };
+
+        await audio.play();
+    };
+
+    run().catch(err => {
+        if (abortController.signal.aborted) return;
+        if (currentRequestId !== requestId) return;
+        notify('error', { message: err?.message || '合成失败' });
+        cleanup();
+    });
+
+    return {
+        stop() {
+            if (currentRequestId === requestId) {
+                stopCurrent();
+            }
+        },
+    };
+}
+
+/**
+ * 停止当前语音（合成中止 + 播放停止 + 资源回收）
+ */
+export function stopCurrent() {
+    if (currentAbort) {
+        try { currentAbort.abort(); } catch { }
+        currentAbort = null;
+    }
+
+    cleanup();
+
+    currentRequestId = null;
+}
+
+// ════════════════════════════════════════════════════════════════════════════
+// 内部
+// ════════════════════════════════════════════════════════════════════════════
+
+function cleanup() {
+    if (currentAudio) {
+        currentAudio.onloadedmetadata = null;
+        currentAudio.onended = null;
+        currentAudio.onerror = null;
+        try { currentAudio.pause(); } catch { }
+        currentAudio = null;
+    }
+
+    if (currentObjectUrl) {
+        URL.revokeObjectURL(currentObjectUrl);
+        currentObjectUrl = null;
+    }
+}
--- a/modules/fourth-wall/fw-voice.js
+++ b/modules/fourth-wall/fw-voice.js
@@ -1,8 +1,7 @@
-// ════════════════════════════════════════════════════════════════════════════
-// 语音模块 - TTS 合成服务
+// ════════════════════════════════════════════════════════════════════════════
+// 语音模块 - 常量与提示词指南
 // ════════════════════════════════════════════════════════════════════════════

-export const TTS_WORKER_URL = 'https://hstts.velure.codes';
 export const DEFAULT_VOICE = 'female_1';
 export const DEFAULT_SPEED = 1.0;

@@ -11,96 +10,9 @@ export const EMOTION_ICONS = {
    happy: '😄', sad: '😢', angry: '😠', surprise: '😮', scare: '😨', hate: '🤢'
 };

-let voiceListCache = null;
-let defaultVoiceKey = DEFAULT_VOICE;
-
-// ════════════════════════════════════════════════════════════════════════════
-// 声音列表管理
-// ════════════════════════════════════════════════════════════════════════════
-
-/**
- * 加载可用声音列表
- */
-export async function loadVoices() {
-    if (voiceListCache) return { voices: voiceListCache, defaultVoice: defaultVoiceKey };
-    
-    try {
-        const res = await fetch(`${TTS_WORKER_URL}/voices`);
-        if (!res.ok) throw new Error(`HTTP ${res.status}`);
-        const data = await res.json();
-        voiceListCache = data.voices || [];
-        defaultVoiceKey = data.defaultVoice || DEFAULT_VOICE;
-        return { voices: voiceListCache, defaultVoice: defaultVoiceKey };
-    } catch (err) {
-        console.error('[FW Voice] 加载声音列表失败:', err);
-        return { voices: [], defaultVoice: DEFAULT_VOICE };
-    }
-}
-
-/**
- * 获取已缓存的声音列表
- */
-export function getVoiceList() {
-    return voiceListCache || [];
-}
-
-/**
- * 获取默认声音
- */
-export function getDefaultVoice() {
-    return defaultVoiceKey;
-}
-
-// ════════════════════════════════════════════════════════════════════════════
-// TTS 合成
-// ════════════════════════════════════════════════════════════════════════════
-
-/**
- * 合成语音
- * @param {string} text - 要合成的文本
- * @param {Object} options - 选项
- * @param {string} [options.voiceKey] - 声音标识
- * @param {number} [options.speed] - 语速 0.5-2.0
- * @param {string} [options.emotion] - 情绪
- * @returns {Promise<string>} base64 编码的音频数据
- */
-export async function synthesizeSpeech(text, options = {}) {
-    const {
-        voiceKey = defaultVoiceKey,
-        speed = DEFAULT_SPEED,
-        emotion = null
-    } = options;
-
-    const requestBody = {
-        voiceKey,
-        text: String(text || ''),
-        speed: Number(speed) || DEFAULT_SPEED,
-        uid: 'xb_' + Date.now(),
-        reqid: crypto.randomUUID?.() || `${Date.now()}_${Math.random().toString(36).slice(2)}`
-    };
-
-    if (emotion && VALID_EMOTIONS.includes(emotion)) {
-        requestBody.emotion = emotion;
-        requestBody.emotionScale = 5;
-    }
-
-    const res = await fetch(TTS_WORKER_URL, {
-        method: 'POST',
-        headers: { 'Content-Type': 'application/json' },
-        body: JSON.stringify(requestBody)
-    });
-
-    if (!res.ok) throw new Error(`TTS HTTP ${res.status}`);
-
-    const data = await res.json();
-    if (data.code !== 3000) throw new Error(data.message || 'TTS 合成失败');
-
-    return data.data; // base64 音频
-}
-
 // ════════════════════════════════════════════════════════════════════════════
 // 提示词指南
-// ════════════════════════════════════════════════════════════════════════════
+// ════════════════════════════════════════════

 export const VOICE_GUIDELINE = `## 模拟语音
 如需发送语音消息，使用以下格式：
--- a/modules/iframe-renderer.js
+++ b/modules/iframe-renderer.js
@@ -358,10 +358,6 @@ export function renderHtmlInIframe(htmlContent, container, preElement) {
        iframe.setAttribute('scrolling', 'no');
        iframe.loading = 'eager';
        
-        if (settings.sandboxMode) {
-            iframe.setAttribute('sandbox', 'allow-scripts');
-        }
-        
        const wrapper = getOrCreateWrapper(preElement);
        wrapper.querySelectorAll('.xiaobaix-iframe').forEach(old => {
            try { old.src = 'about:blank'; } catch (e) {}
--- a/modules/story-outline/story-outline.html
+++ b/modules/story-outline/story-outline.html
--- a/modules/story-summary/generate/llm.js
+++ b/modules/story-summary/generate/llm.js
@@ -236,7 +236,7 @@ All checks passed. Beginning incremental extraction...
 {
  "mindful_prelude":`,

-    userConfirm: `怎么截断了！重新完整生成，只输出JSON，不要任何其他内容
+    userConfirm: `怎么截断了！重新完整生成，只输出JSON，不要任何其他内容，3000字以内
 </Chat_History>`,

    assistantPrefill: JSON_PREFILL
--- a/modules/story-summary/story-summary.html
+++ b/modules/story-summary/story-summary.html
@@ -289,6 +289,33 @@
                    <!-- Trigger Settings -->
                    <div class="settings-section">
                        <div class="settings-section-title">总结设置</div>
+
+                        <!-- Filter Rules -->
+                        <div class="settings-collapse" id="filter-rules-collapse"
+                            style="margin-top:0; margin-bottom: 16px;">
+                            <div class="settings-collapse-header" id="filter-rules-toggle">
+                                <span>文本过滤规则 · <strong id="filter-rules-count">0</strong> 条</span>
+                                <svg class="collapse-icon" viewBox="0 0 24 24" fill="none" stroke="currentColor"
+                                    stroke-width="2">
+                                    <polyline points="6 9 12 15 18 9"></polyline>
+                                </svg>
+                            </div>
+                            <div class="settings-collapse-content hidden" id="filter-rules-content"
+                                style="border-left: 1px solid var(--bdr); border-right: 1px solid var(--bdr); border-bottom: 1px solid var(--bdr); border-radius: 0 0 6px 6px; margin-top: -2px;">
+                                <div class="filter-rules-header">
+                                    <p class="settings-hint" style="margin:0">过滤干扰内容（如思考标签）</p>
+                                    <button class="btn btn-sm btn-add" id="btn-add-filter-rule">
+                                        <svg viewBox="0 0 24 24" width="14" height="14" fill="none"
+                                            stroke="currentColor" stroke-width="2">
+                                            <line x1="12" y1="5" x2="12" y2="19"></line>
+                                            <line x1="5" y1="12" x2="19" y2="12"></line>
+                                        </svg> 添加
+                                    </button>
+                                </div>
+                                <div id="filter-rules-list" class="filter-rules-list"></div>
+                            </div>
+                        </div>
+
                        <div class="settings-row">
                            <div class="settings-field">
                                <label>注入角色</label>
@@ -334,32 +361,6 @@
                            </div>
                        </div>

-                        <!-- Filter Rules -->
-                        <div class="settings-collapse" id="filter-rules-collapse"
-                            style="margin-top:0; margin-bottom: 16px;">
-                            <div class="settings-collapse-header" id="filter-rules-toggle">
-                                <span>文本过滤规则 · <strong id="filter-rules-count">0</strong> 条</span>
-                                <svg class="collapse-icon" viewBox="0 0 24 24" fill="none" stroke="currentColor"
-                                    stroke-width="2">
-                                    <polyline points="6 9 12 15 18 9"></polyline>
-                                </svg>
-                            </div>
-                            <div class="settings-collapse-content hidden" id="filter-rules-content"
-                                style="border-left: 1px solid var(--bdr); border-right: 1px solid var(--bdr); border-bottom: 1px solid var(--bdr); border-radius: 0 0 6px 6px; margin-top: -2px;">
-                                <div class="filter-rules-header">
-                                    <p class="settings-hint" style="margin:0">过滤干扰内容（如思考标签）</p>
-                                    <button class="btn btn-sm btn-add" id="btn-add-filter-rule">
-                                        <svg viewBox="0 0 24 24" width="14" height="14" fill="none"
-                                            stroke="currentColor" stroke-width="2">
-                                            <line x1="12" y1="5" x2="12" y2="19"></line>
-                                            <line x1="5" y1="12" x2="19" y2="12"></line>
-                                        </svg> 添加
-                                    </button>
-                                </div>
-                                <div id="filter-rules-list" class="filter-rules-list"></div>
-                            </div>
-                        </div>
-
                        <!-- Force Insert with wrapper options -->
                        <div class="settings-checkbox-group">
                            <label class="settings-checkbox">
--- a/modules/streaming-generation.js
+++ b/modules/streaming-generation.js
@@ -136,16 +136,11 @@ class StreamingGeneration {
            const opts = { ...baseOptions, ...this.resolveCurrentApiAndModel(baseOptions) };

            const modelLower = String(opts.model || '').toLowerCase();
-            const isClaudeThinkingModel = 
-                modelLower.includes('claude') && 
-                modelLower.includes('thinking') && 
-                !modelLower.includes('nothinking');
+            const isClaudeModel = modelLower.includes('claude');

-            if (isClaudeThinkingModel && Array.isArray(messages) && messages.length > 0) {
-                const lastMsg = messages[messages.length - 1];
-                if (lastMsg?.role === 'assistant') {
-                    console.log('[xbgen] Claude Thinking 模型：移除 assistant prefill');
-                    messages.pop();
+            if (isClaudeModel && Array.isArray(messages) && messages.length > 0) {
+                if (this._convertTrailingAssistantToSystem(messages)) {
+                    console.log('[xbgen] Claude model: converted trailing assistant prefill to system message');
                }
            }

@@ -292,10 +287,14 @@ class StreamingGeneration {
            const logSendRequestError = (err, streamMode) => {
                if (err?.name !== 'AbortError') {
                    const safeApiUrl = String(cmdApiUrl || reverseProxy || oai_settings?.custom_url || '').trim();
+                    const status = this._extractHttpStatus(err);
+                    const isRateLimit = status === 429;
                    try {
                        xbLog.error('streamingGeneration', 'sendRequest failed', {
                            message: err?.message || String(err),
                            name: err?.name,
+                            status,
+                            isRateLimit,
                            stream: !!streamMode,
                            api: String(opts.api || ''),
                            model,
@@ -304,6 +303,12 @@ class StreamingGeneration {
                        });
                    } catch {}
                    console.error('[xbgen:callAPI] sendRequest failed:', err);
+                    if (status) {
+                        console.error(`[xbgen:callAPI] HTTP status=${status} stream=${!!streamMode} api=${String(opts.api || '')} model=${model}`);
+                    }
+                    if (isRateLimit) {
+                        console.error('[xbgen:callAPI] Rate limited (429). Check provider RPM/TPM and concurrent requests.');
+                    }
                }
            };

@@ -470,6 +475,14 @@ class StreamingGeneration {
                return String(session.text || '');
            }

+            const httpStatus = this._extractHttpStatus(err);
+            if (!err?.error && httpStatus) {
+                err.error = {
+                    code: String(httpStatus),
+                    message: String(err?.message || `HTTP ${httpStatus}`),
+                };
+            }
+
            console.error('[StreamingGeneration] Generation error:', err);
            console.error('[StreamingGeneration] error.error =', err?.error);
            try { xbLog.error('streamingGeneration', `processGeneration error sid=${session.id}`, err); } catch {}
@@ -805,6 +818,30 @@ class StreamingGeneration {
        return '';
    }

+    _convertTrailingAssistantToSystem(messages) {
+        if (!Array.isArray(messages) || messages.length < 1) return false;
+        const lastMsg = messages[messages.length - 1];
+        if (lastMsg?.role !== 'assistant') return false;
+
+        const assistantText = this._extractTextFromMessage(lastMsg).trim();
+        const systemText = assistantText ? `Assistant:\n${assistantText}` : 'Assistant:';
+        lastMsg.role = 'system';
+        lastMsg.content = systemText;
+        return true;
+    }
+
+    _extractHttpStatus(err) {
+        const direct = Number(err?.status || err?.statusCode || err?.response?.status);
+        if (Number.isFinite(direct) && direct >= 100 && direct <= 599) return direct;
+        const msg = String(err?.message || '');
+        const m = msg.match(/\bstatus\s+(\d{3})\b/i);
+        if (m) {
+            const s = Number(m[1]);
+            if (Number.isFinite(s)) return s;
+        }
+        return null;
+    }
+
    _getLastMessagesSnapshot() {
        const ctx = getContext();
        const list = Array.isArray(ctx?.chat) ? ctx.chat : [];
--- a/modules/template-editor/template-editor.js
+++ b/modules/template-editor/template-editor.js
@@ -671,8 +671,6 @@ class IframeManager {
    static writeContentToIframe(iframe, content) {
        try {
            const html = buildWrappedHtml(content);
-            const sbox = !!(extension_settings && extension_settings[EXT_ID] && extension_settings[EXT_ID].sandboxMode);
-            if (sbox) iframe.setAttribute('sandbox', 'allow-scripts allow-modals');
            iframe.srcdoc = html;
            const probe = () => {
                const targetOrigin = getIframeTargetOrigin(iframe);
--- a/modules/tts/tts.js
+++ b/modules/tts/tts.js
@@ -1301,6 +1301,7 @@ export async function initTts() {
        openSettings,
        closeSettings,
        player,
+        synthesize: synthesizeForExternal,
        speak: async (text, options = {}) => {
            if (!isModuleEnabled()) return;
            
@@ -1347,6 +1348,106 @@ export async function initTts() {
    };
 }

+// ============ External synthesis API (no enqueue) ============
+
+async function synthesizeForExternal(text, options = {}) {
+    if (!isModuleEnabled()) {
+        throw new Error('TTS 模块未启用');
+    }
+
+    const trimmed = String(text || '').trim();
+    if (!trimmed) {
+        throw new Error('合成文本为空');
+    }
+
+    const { emotion, speaker, signal } = options;
+
+    const mySpeakers = config.volc?.mySpeakers || [];
+    const defaultSpeaker = config.volc?.defaultSpeaker || FREE_DEFAULT_VOICE;
+    const resolved = speaker
+        ? resolveSpeakerWithSource(speaker, mySpeakers, defaultSpeaker)
+        : resolveSpeakerWithSource('', mySpeakers, defaultSpeaker);
+
+    const normalizedEmotion = emotion ? normalizeEmotion(emotion) : '';
+
+    if (resolved.source === 'free') {
+        return await synthesizeFreeBlob(trimmed, resolved.value, normalizedEmotion, signal);
+    }
+
+    if (!isAuthConfigured()) {
+        throw new Error('鉴权音色需要配置 API');
+    }
+
+    return await synthesizeAuthBlob(trimmed, resolved, normalizedEmotion, signal);
+}
+
+async function synthesizeFreeBlob(text, voiceKey, emotion, signal) {
+    const freeSpeed = normalizeSpeed(config?.volc?.speechRate);
+
+    const cacheParams = {
+        providerMode: 'free',
+        text,
+        speaker: voiceKey,
+        freeSpeed,
+        emotion: emotion || '',
+    };
+
+    const cacheHit = await tryLoadLocalCache(cacheParams);
+    if (cacheHit?.entry?.blob) return cacheHit.entry.blob;
+
+    const { synthesizeFreeV1 } = await import('./tts-api.js');
+    const { audioBase64 } = await synthesizeFreeV1({ text, voiceKey, speed: freeSpeed, emotion: emotion || null }, { signal });
+
+    const byteString = atob(audioBase64);
+    const bytes = new Uint8Array(byteString.length);
+    for (let j = 0; j < byteString.length; j++) bytes[j] = byteString.charCodeAt(j);
+    const blob = new Blob([bytes], { type: 'audio/mpeg' });
+
+    const cacheKey = buildCacheKey(cacheParams);
+    storeLocalCache(cacheKey, blob, { text: text.slice(0, 200), textLength: text.length, speaker: voiceKey, resourceId: 'free' }).catch(() => {});
+
+    return blob;
+}
+
+async function synthesizeAuthBlob(text, resolved, emotion, signal) {
+    const resourceId = resolved.resourceId || inferResourceIdBySpeaker(resolved.value);
+    const params = {
+        providerMode: 'auth',
+        appId: config.volc.appId,
+        accessKey: config.volc.accessKey,
+        resourceId,
+        speaker: resolved.value,
+        text,
+        format: 'mp3',
+        sampleRate: 24000,
+        speechRate: speedToV3SpeechRate(config.volc.speechRate),
+        loudnessRate: 0,
+        emotionScale: config.volc.emotionScale,
+        explicitLanguage: config.volc.explicitLanguage,
+        disableMarkdownFilter: config.volc.disableMarkdownFilter,
+        disableEmojiFilter: config.volc.disableEmojiFilter,
+        enableLanguageDetector: config.volc.enableLanguageDetector,
+        maxLengthToFilterParenthesis: config.volc.maxLengthToFilterParenthesis,
+        postProcessPitch: config.volc.postProcessPitch,
+        signal,
+    };
+
+    if (emotion) { params.emotion = emotion; }
+    if (resourceId === 'seed-tts-1.0' && config.volc.useTts11 !== false) { params.model = 'seed-tts-1.1'; }
+    if (config.volc.serverCacheEnabled) { params.cacheConfig = { text_type: 1, use_cache: true }; }
+
+    const cacheHit = await tryLoadLocalCache(params);
+    if (cacheHit?.entry?.blob) return cacheHit.entry.blob;
+
+    const headers = buildV3Headers(resourceId, config);
+    const result = await synthesizeV3(params, headers);
+
+    const cacheKey = buildCacheKey(params);
+    storeLocalCache(cacheKey, result.audioBlob, { text: text.slice(0, 200), textLength: text.length, speaker: resolved.value, resourceId, usage: result.usage || null }).catch(() => {});
+
+    return result.audioBlob;
+}
+
 export function cleanupTts() {
    moduleInitialized = false;
    
--- a/settings.html
+++ b/settings.html
@@ -13,7 +13,6 @@
        <div class="menu-tab" data-target="task" style="border-bottom:1px solid #303030;"><span class="vertical-text">循环任务</span></div>
        <div class="menu-tab" data-target="template" style="border-bottom:1px solid #303030;"><span class="vertical-text">数据互动</span></div>
        <div class="menu-tab" data-target="wallhaven" style="border-bottom:1px solid #303030;"><span class="vertical-text">辅助工具</span></div>
-        <div class="menu-tab" data-target="ena-planner" style="border-bottom:1px solid #303030;"><span class="vertical-text">剧情规划</span></div>
      </div>
      <div class="settings-content">
        <div class="js-memory settings-section" style="display:block;">
@@ -42,10 +41,6 @@
          <div class="section-divider">渲染模式
            <hr class="sysHR" />
          </div>
-          <div class="flex-container">
-            <input type="checkbox" id="xiaobaix_sandbox" />
-            <label for="xiaobaix_sandbox">沙盒模式</label>
-          </div>
          <div class="flex-container">
            <input type="checkbox" id="xiaobaix_use_blob" />
            <label for="xiaobaix_use_blob" class="has-tooltip" data-tooltip="大型html适用">启用Blob渲染</label>
@@ -138,11 +133,6 @@
            </button>
          </div>
        </div>
-        <div class="ena-planner settings-section" style="display:none;">
-          <div id="ena_planner_panel">
-            <!-- Ena Planner UI will be injected here by ena-planner.js -->
-          </div>
-        </div>
        <div class="task settings-section" style="display:none;">
          <div class="section-divider">循环任务
            <hr class="sysHR" />
@@ -530,7 +520,6 @@
 	      storySummary: 'xiaobaix_story_summary_enabled',
 	      tts: 'xiaobaix_tts_enabled',
      storyOutline: 'xiaobaix_story_outline_enabled',
-      sandboxMode: 'xiaobaix_sandbox',
      useBlob: 'xiaobaix_use_blob',
 	      wrapperIframe: 'Wrapperiframe',
 	      renderEnabled: 'xiaobaix_render_enabled',
@@ -547,9 +536,8 @@
      if (el) { el.checked = !!enabled; try { $(el).trigger('change'); } catch (e) { } }
    }
    function captureStates() {
-      const out = { modules: {}, sandboxMode: false, useBlob: false, wrapperIframe: false, renderEnabled: true };
+      const out = { modules: {}, useBlob: false, wrapperIframe: false, renderEnabled: true };
      try { MODULE_KEYS.forEach(k => { out.modules[k] = !!(extension_settings[EXT_ID][k] && extension_settings[EXT_ID][k].enabled); }); } catch (e) { }
-      try { out.sandboxMode = !!extension_settings[EXT_ID].sandboxMode; } catch (e) { }
      try { out.useBlob = !!extension_settings[EXT_ID].useBlob; } catch (e) { }
      try { out.wrapperIframe = !!extension_settings[EXT_ID].wrapperIframe; } catch (e) { }
      try { out.renderEnabled = extension_settings[EXT_ID].renderEnabled !== false; } catch (e) { }
@@ -558,10 +546,6 @@
    function applyStates(st) {
      if (!st) return;
      try { Object.keys(st.modules || {}).forEach(k => setModuleEnabled(k, !!st.modules[k])); } catch (e) { }
-      try {
-        extension_settings[EXT_ID].sandboxMode = !!st.sandboxMode;
-        const el = $id('xiaobaix_sandbox'); if (el) { el.checked = !!st.sandboxMode; if (window.isXiaobaixEnabled) try { $(el).trigger('change'); } catch (e) { } }
-      } catch (e) { }
      try {
        extension_settings[EXT_ID].useBlob = !!st.useBlob;
        const el = $id('xiaobaix_use_blob'); if (el) { el.checked = !!st.useBlob; if (window.isXiaobaixEnabled) try { $(el).trigger('change'); } catch (e) { } }
@@ -579,10 +563,6 @@
    function applyResetDefaults() {
      DEFAULTS_ON.forEach(k => setModuleEnabled(k, true));
      DEFAULTS_OFF.forEach(k => setModuleEnabled(k, false));
-      try {
-        extension_settings[EXT_ID].sandboxMode = false; const sb = $id(KEY_TO_CHECKBOX.sandboxMode);
-        if (sb) { sb.checked = false; try { $(sb).trigger('change'); } catch (e) { } }
-      } catch (e) { }
      try {
        extension_settings[EXT_ID].useBlob = false; const bl = $id(KEY_TO_CHECKBOX.useBlob);
        if (bl) { bl.checked = false; try { $(bl).trigger('change'); } catch (e) { } }
@@ -788,4 +768,3 @@
    <div class="cloud-task-intro" style="color:#888;font-size:.9em;text-align:left;"></div>
  </div>
 </div>
-