调整1.0鉴权音色方式

2026-01-20 10:36:35 +08:00
parent dc507d7d6d
commit d1a4aca025
3 changed files with 93 additions and 73 deletions
--- a/modules/tts/tts-auth-provider.js
+++ b/modules/tts/tts-auth-provider.js
@@ -40,7 +40,10 @@ export function speedToV3SpeechRate(speed) {
    return Math.round((normalizeSpeed(speed) - 1) * 100);
 }

-export function inferResourceIdBySpeaker(value) {
+export function inferResourceIdBySpeaker(value, explicitResourceId = null) {
+    if (explicitResourceId) {
+        return explicitResourceId;
+    }
    const v = (value || '').trim();
    const lower = v.toLowerCase();
    if (lower.startsWith('icl_') || lower.startsWith('s_')) {
@@ -110,7 +113,7 @@ export async function speakSegmentAuth(messageId, segment, segmentIndex, batchId
    } = ctx;
    
    const speaker = segment.resolvedSpeaker;
-    const resourceId = inferResourceIdBySpeaker(speaker);
+    const resourceId = segment.resolvedResourceId || inferResourceIdBySpeaker(speaker);
    const params = buildSynthesizeParams({ text: segment.text, speaker, resourceId }, config);
    const emotion = normalizeEmotion(segment.emotion);
    const contextTexts = resolveContextTexts(segment.context, resourceId);
@@ -171,7 +174,7 @@ export async function speakSegmentAuth(messageId, segment, segmentIndex, batchId
 async function playWithStreaming(messageId, segment, segmentIndex, batchId, params, headers, ctx) {
    const { player, storeLocalCache, buildCacheKey, updateState } = ctx;
    const speaker = segment.resolvedSpeaker;
-    const resourceId = inferResourceIdBySpeaker(speaker);
+    const resourceId = params.resourceId;
    
    const controller = new AbortController();
    const chunks = [];
@@ -250,7 +253,7 @@ async function playWithStreaming(messageId, segment, segmentIndex, batchId, para
 async function playWithoutStreaming(messageId, segment, segmentIndex, batchId, params, headers, ctx) {
    const { player, storeLocalCache, buildCacheKey, updateState } = ctx;
    const speaker = segment.resolvedSpeaker;
-    const resourceId = inferResourceIdBySpeaker(speaker);
+    const resourceId = params.resourceId;
    
    const result = await synthesizeV3(params, headers);
    updateState({ audioBlob: result.audioBlob, usage: result.usage, status: 'queued' });
--- a/modules/tts/tts-overlay.html
+++ b/modules/tts/tts-overlay.html
@@ -1408,13 +1408,20 @@ select.input { cursor: pointer; }
                                    <label class="form-label">音色 ID</label>
                                    <input type="text" id="newVoiceId" class="input" placeholder="如 S_xxx">
                                </div>
-                                <div class="form-group">
-                                    <label class="form-label">名称</label>
-                                    <input type="text" id="newVoiceName" class="input" placeholder="显示名称">
-                                </div>
-                                <button class="btn btn-primary" id="addMySpeakerBtn" style="margin-top: 18px;"><i class="fa-solid fa-plus"></i></button>
-                            </div>
-                        </div>
+                                <div class="form-group">
+                                    <label class="form-label">名称</label>
+                                    <input type="text" id="newVoiceName" class="input" placeholder="显示名称">
+                                </div>
+                                <div class="form-group">
+                                    <label class="form-label">复刻版本</label>
+                                    <select id="newVoiceResourceId" class="input">
+                                        <option value="seed-icl-2.0">复刻 2.0</option>
+                                        <option value="seed-icl-1.0">复刻 1.0</option>
+                                    </select>
+                                </div>
+                                <button class="btn btn-primary" id="addMySpeakerBtn" style="margin-top: 18px;"><i class="fa-solid fa-plus"></i></button>
+                            </div>
+                        </div>
                    </div>
                </div>

@@ -2149,14 +2156,15 @@ function renderAuthVoiceList() {
 // 数据处理
 // ═══════════════════════════════════════════════════════════════════════════

-function normalizeMySpeakers(list) {
-    if (!Array.isArray(list)) return [];
-    return list.map(item => ({
-        name: String(item?.name || '').trim(),
-        value: String(item?.value || '').trim(),
-        source: item?.source || getVoiceSource(item?.value || ''),
-    })).filter(item => item.value);
-}
+function normalizeMySpeakers(list) {
+    if (!Array.isArray(list)) return [];
+    return list.map(item => ({
+        name: String(item?.name || '').trim(),
+        value: String(item?.value || '').trim(),
+        source: item?.source || getVoiceSource(item?.value || ''),
+        resourceId: item?.resourceId || null,
+    })).filter(item => item.value);
+}

 function applyCacheStats(stats = {}) {
    $('cacheCount').textContent = stats.count ?? 0;
@@ -2265,13 +2273,16 @@ function doTestVoice(speaker, source, textElId, statusElId) {
    
    setTestStatus(statusElId, 'playing', '正在合成...');
    
-    post('xb-tts:test-speak', {
+    const speakerItem = mySpeakers.find(s => s.value === speaker);
+    const resolvedResourceId = speakerItem?.resourceId;
+
+    post('xb-tts:test-speak', {
        text,
        speaker,
        source,
-        resourceId: source === 'auth' ? inferResourceIdBySpeaker(speaker) : '',
-    });
-}
+        resourceId: source === 'auth' ? (resolvedResourceId || inferResourceIdBySpeaker(speaker)) : '',
+    });
+}

 // ═══════════════════════════════════════════════════════════════════════════
 // 消息处理
@@ -2434,17 +2445,18 @@ document.addEventListener('DOMContentLoaded', () => {
        post('xb-tts:toast', { type: 'success', message: `已添加：${name}` });
    });

-    $('addMySpeakerBtn').addEventListener('click', () => {
-        const id = $('newVoiceId').value.trim();
-        const name = $('newVoiceName').value.trim();
-        if (!id) { post('xb-tts:toast', { type: 'error', message: '请输入音色ID' }); return; }
-        
-        if (!isInMyList(id)) {
-            mySpeakers.push({ name: name || id, value: id, source: 'auth' });
-        }
-        selectedVoiceValue = id;
-        $('newVoiceId').value = '';
-        $('newVoiceName').value = '';
+    $('addMySpeakerBtn').addEventListener('click', () => {
+        const id = $('newVoiceId').value.trim();
+        const name = $('newVoiceName').value.trim();
+        const resourceId = $('newVoiceResourceId').value;
+        if (!id) { post('xb-tts:toast', { type: 'error', message: '请输入音色ID' }); return; }
+        
+        if (!isInMyList(id)) {
+            mySpeakers.push({ name: name || id, value: id, source: 'auth', resourceId });
+        }
+        selectedVoiceValue = id;
+        $('newVoiceId').value = '';
+        $('newVoiceName').value = '';
        
        renderMyVoiceList();
        updateCurrentVoiceDisplay();
--- a/modules/tts/tts.js
+++ b/modules/tts/tts.js
@@ -250,47 +250,51 @@ function resolveSpeakerWithSource(speakerName, mySpeakers, defaultSpeaker) {
        });
    }
    
-    if (!speakerName) {
-        const defaultItem = list.find(s => s.value === defaultSpeaker);
-        return {
-            value: defaultSpeaker,
-            source: defaultItem?.source || getVoiceSource(defaultSpeaker)
-        };
-    }
+    if (!speakerName) {
+        const defaultItem = list.find(s => s.value === defaultSpeaker);
+        return {
+            value: defaultSpeaker,
+            source: defaultItem?.source || getVoiceSource(defaultSpeaker),
+            resourceId: defaultItem?.resourceId || null
+        };
+    }
    
    const byName = list.find(s => s.name === speakerName);
    console.log('[TTS Debug] byName 查找结果:', byName); // ★ 调试
    
-    if (byName?.value) {
-        return {
-            value: byName.value,
-            source: byName.source || getVoiceSource(byName.value)
-        };
-    }
+    if (byName?.value) {
+        return {
+            value: byName.value,
+            source: byName.source || getVoiceSource(byName.value),
+            resourceId: byName.resourceId || null
+        };
+    }
    
    const byValue = list.find(s => s.value === speakerName);
    console.log('[TTS Debug] byValue 查找结果:', byValue); // ★ 调试
    
-    if (byValue?.value) {
-        return {
-            value: byValue.value,
-            source: byValue.source || getVoiceSource(byValue.value)
-        };
-    }
+    if (byValue?.value) {
+        return {
+            value: byValue.value,
+            source: byValue.source || getVoiceSource(byValue.value),
+            resourceId: byValue.resourceId || null
+        };
+    }
    
-    if (FREE_VOICE_KEYS.has(speakerName)) {
-        return { value: speakerName, source: 'free' };
-    }
+    if (FREE_VOICE_KEYS.has(speakerName)) {
+        return { value: speakerName, source: 'free', resourceId: null };
+    }
    
    // ★ 回退到默认，这是问题发生的地方
    console.warn('[TTS Debug] 未找到匹配音色，回退到默认:', defaultSpeaker);
    
-    const defaultItem = list.find(s => s.value === defaultSpeaker);
-    return {
-        value: defaultSpeaker,
-        source: defaultItem?.source || getVoiceSource(defaultSpeaker)
-    };
-}
+    const defaultItem = list.find(s => s.value === defaultSpeaker);
+    return {
+        value: defaultSpeaker,
+        source: defaultItem?.source || getVoiceSource(defaultSpeaker),
+        resourceId: defaultItem?.resourceId || null
+    };
+}

 // ============ 缓存管理 ============

@@ -616,16 +620,17 @@ async function speakMessage(messageId, { mode = 'manual' } = {}) {
        return;
    }

-    const resolvedSegments = segments.map(seg => {
-        const resolved = seg.speaker 
-            ? resolveSpeakerWithSource(seg.speaker, mySpeakers, defaultSpeaker)
-            : defaultResolved;
-        return { 
-            ...seg, 
-            resolvedSpeaker: resolved.value, 
-            resolvedSource: resolved.source 
-        };
-    });
+    const resolvedSegments = segments.map(seg => {
+        const resolved = seg.speaker 
+            ? resolveSpeakerWithSource(seg.speaker, mySpeakers, defaultSpeaker)
+            : defaultResolved;
+        return { 
+            ...seg, 
+            resolvedSpeaker: resolved.value, 
+            resolvedSource: resolved.source,
+            resolvedResourceId: resolved.resourceId
+        };
+    });

    const needsAuth = resolvedSegments.some(s => s.resolvedSource === 'auth');
    if (needsAuth && !isAuthConfigured()) {
@@ -1325,7 +1330,7 @@ export async function initTts() {
                    return;
                }
                
-                const resourceId = options.resourceId || inferResourceIdBySpeaker(resolved.value);
+                const resourceId = options.resourceId || resolved.resourceId || inferResourceIdBySpeaker(resolved.value);
                const result = await synthesizeV3({
                    appId: config.volc.appId,
                    accessKey: config.volc.accessKey,