Merge branch 'master' of https://gitea.xiaowang.eu.org/wx44wx/AI-VideoAssistant

api/app/models.py

@@ -38,6 +38,17 @@ class Voice(Base):
     user = relationship("User", foreign_keys=[user_id])
 
 
+class VendorCredential(Base):
+    __tablename__ = "vendor_credentials"
+
+    vendor_key: Mapped[str] = mapped_column(String(64), primary_key=True)
+    vendor_name: Mapped[str] = mapped_column(String(128), nullable=False)
+    api_key: Mapped[str] = mapped_column(String(512), nullable=False)
+    base_url: Mapped[Optional[str]] = mapped_column(String(512), nullable=True)
+    created_at: Mapped[datetime] = mapped_column(DateTime, default=datetime.utcnow)
+    updated_at: Mapped[datetime] = mapped_column(DateTime, default=datetime.utcnow)
+
+
 # ============ LLM Model ============
 class LLMModel(Base):
     __tablename__ = "llm_models"

api/app/routers/knowledge.py

@@ -146,12 +146,27 @@ def delete_knowledge_base(kb_id: str, db: Session = Depends(get_db)):
 @router.post("/bases/{kb_id}/documents")
 def upload_document(
     kb_id: str,
-    data: KnowledgeDocumentCreate,
+    data: Optional[KnowledgeDocumentCreate] = None,
+    name: Optional[str] = Query(default=None),
+    size: Optional[str] = Query(default=None),
+    file_type: Optional[str] = Query(default=None),
+    storage_url: Optional[str] = Query(default=None),
     db: Session = Depends(get_db)
 ):
     kb = db.query(KnowledgeBase).filter(KnowledgeBase.id == kb_id).first()
     if not kb:
         raise HTTPException(status_code=404, detail="Knowledge base not found")
+
+    if data is None:
+        if not name or not size:
+            raise HTTPException(status_code=422, detail="name and size are required")
+        data = KnowledgeDocumentCreate(
+            name=name,
+            size=size,
+            fileType=file_type or "txt",
+            storageUrl=storage_url,
+        )
+
     doc = KnowledgeDocument(
         id=str(uuid.uuid4())[:8],
         kb_id=kb_id,

api/app/routers/voices.py

@@ -1,14 +1,65 @@
+import base64
+import os
+import uuid
+from datetime import datetime
+from typing import Optional
+
+import httpx
 from fastapi import APIRouter, Depends, HTTPException
 from sqlalchemy.orm import Session
-from typing import Optional
-import uuid
 
 from ..db import get_db
-from ..models import Voice
+from ..models import VendorCredential, Voice
-from ..schemas import VoiceCreate, VoiceUpdate, VoiceOut
+from ..schemas import (
+    VendorCredentialOut,
+    VendorCredentialUpsert,
+    VoiceCreate,
+    VoiceOut,
+    VoicePreviewRequest,
+    VoicePreviewResponse,
+    VoiceUpdate,
+)
 
 router = APIRouter(prefix="/voices", tags=["Voices"])
 
+SILICONFLOW_DEFAULT_MODEL = "FunAudioLLM/CosyVoice2-0.5B"
+
+
+def _is_siliconflow_vendor(vendor: str) -> bool:
+    return vendor.strip().lower() in {"siliconflow", "硅基流动"}
+
+
+def _canonical_vendor_key(vendor: str) -> str:
+    normalized = vendor.strip().lower()
+    alias_map = {
+        "硅基流动": "siliconflow",
+        "siliconflow": "siliconflow",
+        "ali": "ali",
+        "volcano": "volcano",
+        "minimax": "minimax",
+    }
+    return alias_map.get(normalized, normalized)
+
+
+def _default_tts_base_url(vendor_key: str) -> Optional[str]:
+    defaults = {
+        "siliconflow": "https://api.siliconflow.cn/v1",
+    }
+    return defaults.get(vendor_key)
+
+
+def _resolve_vendor_credential(db: Session, vendor: str) -> Optional[VendorCredential]:
+    vendor_key = _canonical_vendor_key(vendor)
+    return db.query(VendorCredential).filter(VendorCredential.vendor_key == vendor_key).first()
+
+
+def _build_siliconflow_voice_key(voice: Voice, model: str) -> str:
+    if voice.voice_key:
+        return voice.voice_key
+    if ":" in voice.id:
+        return voice.id
+    return f"{model}:{voice.id}"
+
 
 @router.get("")
 def list_voices(
@@ -37,16 +88,26 @@ def list_voices(
 @router.post("", response_model=VoiceOut)
 def create_voice(data: VoiceCreate, db: Session = Depends(get_db)):
     """创建声音"""
+    vendor = data.vendor.strip()
+    model = data.model
+    voice_key = data.voice_key
+
+    if _is_siliconflow_vendor(vendor):
+        model = model or SILICONFLOW_DEFAULT_MODEL
+        if not voice_key:
+            raw_id = (data.id or data.name).strip()
+            voice_key = raw_id if ":" in raw_id else f"{model}:{raw_id}"
+
     voice = Voice(
         id=data.id or str(uuid.uuid4())[:8],
         user_id=1,
         name=data.name,
-        vendor=data.vendor,
+        vendor=vendor,
         gender=data.gender,
         language=data.language,
         description=data.description,
-        model=data.model,
+        model=model,
-        voice_key=data.voice_key,
+        voice_key=voice_key,
         speed=data.speed,
         gain=data.gain,
         pitch=data.pitch,
@@ -75,6 +136,16 @@ def update_voice(id: str, data: VoiceUpdate, db: Session = Depends(get_db)):
         raise HTTPException(status_code=404, detail="Voice not found")
 
     update_data = data.model_dump(exclude_unset=True)
+    if "vendor" in update_data and update_data["vendor"] is not None:
+        update_data["vendor"] = update_data["vendor"].strip()
+
+    vendor_for_defaults = update_data.get("vendor", voice.vendor)
+    if _is_siliconflow_vendor(vendor_for_defaults):
+        model = update_data.get("model") or voice.model or SILICONFLOW_DEFAULT_MODEL
+        voice_key = update_data.get("voice_key") or voice.voice_key
+        update_data["model"] = model
+        update_data["voice_key"] = voice_key or _build_siliconflow_voice_key(voice, model)
+
     for field, value in update_data.items():
         setattr(voice, field, value)
 
@@ -92,3 +163,111 @@ def delete_voice(id: str, db: Session = Depends(get_db)):
     db.delete(voice)
     db.commit()
     return {"message": "Deleted successfully"}
+
+
+@router.get("/vendors/credentials")
+def list_vendor_credentials(db: Session = Depends(get_db)):
+    items = db.query(VendorCredential).order_by(VendorCredential.updated_at.desc()).all()
+    return {"list": items, "total": len(items)}
+
+
+@router.get("/vendors/credentials/{vendor_key}", response_model=VendorCredentialOut)
+def get_vendor_credential(vendor_key: str, db: Session = Depends(get_db)):
+    key = _canonical_vendor_key(vendor_key)
+    item = db.query(VendorCredential).filter(VendorCredential.vendor_key == key).first()
+    if not item:
+        raise HTTPException(status_code=404, detail="Vendor credential not found")
+    return item
+
+
+@router.put("/vendors/credentials/{vendor_key}", response_model=VendorCredentialOut)
+def upsert_vendor_credential(vendor_key: str, data: VendorCredentialUpsert, db: Session = Depends(get_db)):
+    key = _canonical_vendor_key(vendor_key)
+    item = db.query(VendorCredential).filter(VendorCredential.vendor_key == key).first()
+
+    if item:
+        item.vendor_name = data.vendor_name or item.vendor_name
+        item.api_key = data.api_key
+        item.base_url = data.base_url
+        item.updated_at = datetime.utcnow()
+    else:
+        item = VendorCredential(
+            vendor_key=key,
+            vendor_name=data.vendor_name or vendor_key,
+            api_key=data.api_key,
+            base_url=data.base_url,
+        )
+        db.add(item)
+
+    db.commit()
+    db.refresh(item)
+    return item
+
+
+@router.delete("/vendors/credentials/{vendor_key}")
+def delete_vendor_credential(vendor_key: str, db: Session = Depends(get_db)):
+    key = _canonical_vendor_key(vendor_key)
+    item = db.query(VendorCredential).filter(VendorCredential.vendor_key == key).first()
+    if not item:
+        raise HTTPException(status_code=404, detail="Vendor credential not found")
+    db.delete(item)
+    db.commit()
+    return {"message": "Deleted successfully"}
+
+
+@router.post("/{id}/preview", response_model=VoicePreviewResponse)
+def preview_voice(id: str, data: VoicePreviewRequest, db: Session = Depends(get_db)):
+    """试听指定声音，基于 OpenAI-compatible /audio/speech 接口。"""
+    voice = db.query(Voice).filter(Voice.id == id).first()
+    if not voice:
+        raise HTTPException(status_code=404, detail="Voice not found")
+
+    text = data.text.strip()
+    if not text:
+        raise HTTPException(status_code=400, detail="Preview text cannot be empty")
+
+    credential = _resolve_vendor_credential(db, voice.vendor)
+    api_key = (data.api_key or "").strip()
+    if not api_key and credential:
+        api_key = credential.api_key
+    if not api_key:
+        api_key = os.getenv("SILICONFLOW_API_KEY") if _is_siliconflow_vendor(voice.vendor) else ""
+    if not api_key:
+        raise HTTPException(status_code=400, detail=f"Vendor API key is required for {voice.vendor}")
+
+    model = voice.model or SILICONFLOW_DEFAULT_MODEL
+    vendor_key = _canonical_vendor_key(voice.vendor)
+    base_url = (credential.base_url.strip() if credential and credential.base_url else "") or _default_tts_base_url(vendor_key)
+    if not base_url:
+        raise HTTPException(status_code=400, detail=f"Vendor base_url is required for {voice.vendor}")
+    tts_api_url = f"{base_url.rstrip('/')}/audio/speech"
+
+    payload = {
+        "model": model,
+        "input": text,
+        "voice": voice.voice_key or _build_siliconflow_voice_key(voice, model),
+        "response_format": "mp3",
+        "speed": data.speed if data.speed is not None else voice.speed,
+    }
+
+    try:
+        with httpx.Client(timeout=45.0) as client:
+            response = client.post(
+                tts_api_url,
+                headers={"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"},
+                json=payload,
+            )
+    except Exception as exc:
+        raise HTTPException(status_code=502, detail=f"TTS request failed: {exc}") from exc
+
+    if response.status_code != 200:
+        detail = response.text
+        try:
+            detail_json = response.json()
+            detail = detail_json.get("error", {}).get("message") or detail_json.get("detail") or detail
+        except Exception:
+            pass
+        raise HTTPException(status_code=502, detail=f"TTS vendor error: {detail}")
+
+    audio_base64 = base64.b64encode(response.content).decode("utf-8")
+    return VoicePreviewResponse(success=True, audio_url=f"data:audio/mpeg;base64,{audio_base64}")

api/app/routers/workflows.py

@@ -1,6 +1,5 @@
 from fastapi import APIRouter, Depends, HTTPException
 from sqlalchemy.orm import Session
-from typing import List
 import uuid
 from datetime import datetime
 
@@ -11,7 +10,7 @@ from ..schemas import WorkflowCreate, WorkflowUpdate, WorkflowOut
 router = APIRouter(prefix="/workflows", tags=["Workflows"])
 
 
-@router.get("", response_model=List[WorkflowOut])
+@router.get("")
 def list_workflows(
     page: int = 1,
     limit: int = 50,
@@ -62,8 +61,12 @@ def update_workflow(id: str, data: WorkflowUpdate, db: Session = Depends(get_db)
         raise HTTPException(status_code=404, detail="Workflow not found")
 
     update_data = data.model_dump(exclude_unset=True)
+    field_map = {
+        "nodeCount": "node_count",
+        "globalPrompt": "global_prompt",
+    }
     for field, value in update_data.items():
-        setattr(workflow, field, value)
+        setattr(workflow, field_map.get(field, field), value)
 
     workflow.updated_at = datetime.utcnow().isoformat()
     db.commit()

api/app/schemas.py

@@ -61,6 +61,9 @@ class VoiceCreate(VoiceBase):
 
 class VoiceUpdate(BaseModel):
     name: Optional[str] = None
+    vendor: Optional[str] = None
+    gender: Optional[str] = None
+    language: Optional[str] = None
     description: Optional[str] = None
     model: Optional[str] = None
     voice_key: Optional[str] = None
@@ -88,6 +91,7 @@ class VoiceOut(VoiceBase):
 
 class VoicePreviewRequest(BaseModel):
     text: str
+    api_key: Optional[str] = None
     speed: Optional[float] = None
     gain: Optional[int] = None
     pitch: Optional[int] = None
@@ -100,6 +104,24 @@ class VoicePreviewResponse(BaseModel):
     error: Optional[str] = None
 
 
+class VendorCredentialUpsert(BaseModel):
+    vendor_name: Optional[str] = None
+    api_key: str
+    base_url: Optional[str] = None
+
+
+class VendorCredentialOut(BaseModel):
+    vendor_key: str
+    vendor_name: str
+    api_key: str
+    base_url: Optional[str] = None
+    created_at: Optional[datetime] = None
+    updated_at: Optional[datetime] = None
+
+    class Config:
+        from_attributes = True
+
+
 # ============ LLM Model ============
 class LLMModelBase(BaseModel):
     name: str

api/tests/test_voices.py

@@ -1,4 +1,5 @@
 """Tests for Voice API endpoints"""
+import base64
 import pytest
 
 
@@ -130,3 +131,110 @@ class TestVoiceAPI:
         data = response.json()
         for voice in data["list"]:
             assert voice["gender"] == "Female"
+
+    def test_preview_voice_success(self, client, monkeypatch):
+        """Test preview voice endpoint returns audio data URL"""
+        from app.routers import voices as voice_router
+
+        class DummyResponse:
+            status_code = 200
+            content = b"fake-mp3-bytes"
+            text = "ok"
+
+            def json(self):
+                return {}
+
+        class DummyClient:
+            def __init__(self, *args, **kwargs):
+                pass
+
+            def __enter__(self):
+                return self
+
+            def __exit__(self, exc_type, exc, tb):
+                return False
+
+            def post(self, *args, **kwargs):
+                return DummyResponse()
+
+        monkeypatch.setenv("SILICONFLOW_API_KEY", "test-key")
+        monkeypatch.setattr(voice_router.httpx, "Client", DummyClient)
+
+        create_resp = client.post("/api/voices", json={
+            "id": "anna",
+            "name": "Anna",
+            "vendor": "SiliconFlow",
+            "gender": "Female",
+            "language": "zh",
+            "description": "system voice",
+            "model": "FunAudioLLM/CosyVoice2-0.5B",
+            "voice_key": "FunAudioLLM/CosyVoice2-0.5B:anna"
+        })
+        assert create_resp.status_code == 200
+
+        preview_resp = client.post("/api/voices/anna/preview", json={"text": "你好"})
+        assert preview_resp.status_code == 200
+        payload = preview_resp.json()
+        assert payload["success"] is True
+        assert payload["audio_url"].startswith("data:audio/mpeg;base64,")
+        encoded = payload["audio_url"].split(",", 1)[1]
+        assert base64.b64decode(encoded) == b"fake-mp3-bytes"
+
+    def test_vendor_credential_persist_and_preview_use_db_key(self, client, monkeypatch):
+        """Test vendor credential persisted in DB and used by preview endpoint"""
+        from app.routers import voices as voice_router
+
+        captured_auth = {"value": ""}
+
+        class DummyResponse:
+            status_code = 200
+            content = b"fake-mp3"
+            text = "ok"
+
+            def json(self):
+                return {}
+
+        class DummyClient:
+            def __init__(self, *args, **kwargs):
+                pass
+
+            def __enter__(self):
+                return self
+
+            def __exit__(self, exc_type, exc, tb):
+                return False
+
+            def post(self, *args, **kwargs):
+                headers = kwargs.get("headers", {})
+                captured_auth["value"] = headers.get("Authorization", "")
+                return DummyResponse()
+
+        monkeypatch.delenv("SILICONFLOW_API_KEY", raising=False)
+        monkeypatch.setattr(voice_router.httpx, "Client", DummyClient)
+
+        save_cred = client.put(
+            "/api/voices/vendors/credentials/siliconflow",
+            json={
+                "vendor_name": "SiliconFlow",
+                "api_key": "db-key-123",
+                "base_url": "https://api.siliconflow.cn/v1"
+            },
+        )
+        assert save_cred.status_code == 200
+        assert save_cred.json()["vendor_key"] == "siliconflow"
+
+        create_resp = client.post("/api/voices", json={
+            "id": "anna2",
+            "name": "Anna 2",
+            "vendor": "SiliconFlow",
+            "gender": "Female",
+            "language": "zh",
+            "description": "voice",
+            "model": "FunAudioLLM/CosyVoice2-0.5B",
+            "voice_key": "FunAudioLLM/CosyVoice2-0.5B:anna"
+        })
+        assert create_resp.status_code == 200
+
+        preview_resp = client.post("/api/voices/anna2/preview", json={"text": "hello"})
+        assert preview_resp.status_code == 200
+        assert captured_auth["value"] == "Bearer db-key-123"

web/README.md

@@ -16,6 +16,6 @@ View your app in AI Studio: https://ai.studio/apps/drive/1Cg9WH_2bOQEHVVj-lSN5l2
 1. Install dependencies:
    `npm install`
 2. Set the `GEMINI_API_KEY` in [.env.local](.env.local) to your Gemini API key
-3. Optional: set `VITE_API_BASE_URL` (for backend API, default `http://localhost:8000/api`)
+3. Optional: set `VITE_API_BASE_URL` (for backend API, default `http://127.0.0.1:8100/api`)
 4. Run the app:
    `npm run dev`

web/pages/VoiceLibrary.tsx

@@ -1,40 +1,65 @@
 
 import React, { useEffect, useState, useRef } from 'react';
-import { Search, Mic2, Play, Pause, Upload, X, Filter, Plus, Volume2, Sparkles, Wand2, ChevronDown } from 'lucide-react';
+import { Search, Mic2, Play, Pause, Upload, X, Filter, Plus, Volume2, Sparkles, Wand2, ChevronDown, Pencil, Trash2 } from 'lucide-react';
 import { Button, Input, TableHeader, TableRow, TableHead, TableCell, Dialog, Badge } from '../components/UI';
-import { mockVoices } from '../services/mockData';
+import { VendorCredential, Voice } from '../types';
-import { Voice } from '../types';
+import { createVoice, deleteVoice, fetchVendorCredentials, fetchVoices, previewVoice, saveVendorCredential, updateVoice } from '../services/backendApi';
-import { fetchVoices } from '../services/backendApi';
+
+const VENDOR_OPTIONS = [
+  { key: 'siliconflow', label: '硅基流动 (SiliconFlow)' },
+  { key: 'ali', label: 'Ali' },
+  { key: 'volcano', label: 'Volcano' },
+  { key: 'minimax', label: 'Minimax' },
+];
 
 export const VoiceLibraryPage: React.FC = () => {
   const [voices, setVoices] = useState<Voice[]>([]);
   const [searchTerm, setSearchTerm] = useState('');
-  const [vendorFilter, setVendorFilter] = useState<'all' | 'Ali' | 'Volcano' | 'Minimax' | '硅基流动'>('all');
+  const [vendorFilter, setVendorFilter] = useState<'all' | 'Ali' | 'Volcano' | 'Minimax' | '硅基流动' | 'SiliconFlow'>('all');
   const [genderFilter, setGenderFilter] = useState<'all' | 'Male' | 'Female'>('all');
   const [langFilter, setLangFilter] = useState<'all' | 'zh' | 'en'>('all');
   
   const [playingVoiceId, setPlayingVoiceId] = useState<string | null>(null);
   const [isCloneModalOpen, setIsCloneModalOpen] = useState(false);
   const [isAddModalOpen, setIsAddModalOpen] = useState(false);
+  const [editingVoice, setEditingVoice] = useState<Voice | null>(null);
   const [isLoading, setIsLoading] = useState(true);
+  const [playLoadingId, setPlayLoadingId] = useState<string | null>(null);
+  const [vendorCredentials, setVendorCredentials] = useState<Record<string, VendorCredential>>({});
+  const [credentialVendorKey, setCredentialVendorKey] = useState('siliconflow');
+  const [credentialApiKey, setCredentialApiKey] = useState('');
+  const [credentialBaseUrl, setCredentialBaseUrl] = useState('');
+  const [isSavingCredential, setIsSavingCredential] = useState(false);
+  const audioRef = useRef<HTMLAudioElement | null>(null);
 
   useEffect(() => {
-    const loadVoices = async () => {
+    const loadVoicesAndCredentials = async () => {
       setIsLoading(true);
       try {
-        const list = await fetchVoices();
+        const [list, credentials] = await Promise.all([fetchVoices(), fetchVendorCredentials()]);
-        setVoices(list.length > 0 ? list : mockVoices);
+        setVoices(list);
+        const mapped = credentials.reduce((acc, item) => {
+          acc[item.vendorKey] = item;
+          return acc;
+        }, {} as Record<string, VendorCredential>);
+        setVendorCredentials(mapped);
       } catch (error) {
         console.error(error);
-        setVoices(mockVoices);
+        setVoices([]);
       } finally {
         setIsLoading(false);
       }
     };
 
-    loadVoices();
+    loadVoicesAndCredentials();
   }, []);
 
+  useEffect(() => {
+    const selected = vendorCredentials[credentialVendorKey];
+    setCredentialApiKey(selected?.apiKey || '');
+    setCredentialBaseUrl(selected?.baseUrl || '');
+  }, [credentialVendorKey, vendorCredentials]);
+
   const filteredVoices = voices.filter(voice => {
     const matchesSearch = voice.name.toLowerCase().includes(searchTerm.toLowerCase());
     const matchesVendor = vendorFilter === 'all' || voice.vendor === vendorFilter;
@@ -43,23 +68,80 @@ export const VoiceLibraryPage: React.FC = () => {
     return matchesSearch && matchesVendor && matchesGender && matchesLang;
   });
 
-  const handlePlayToggle = (id: string) => {
+  const handlePlayToggle = async (voice: Voice) => {
-    if (playingVoiceId === id) {
+    if (playingVoiceId === voice.id && audioRef.current) {
+      audioRef.current.pause();
+      audioRef.current.currentTime = 0;
       setPlayingVoiceId(null);
-    } else {
+      return;
-      setPlayingVoiceId(id);
+    }
-      setTimeout(() => {
+    try {
-          setPlayingVoiceId((current) => current === id ? null : current);
+      setPlayLoadingId(voice.id);
-      }, 3000);
+      const audioUrl = await previewVoice(
+        voice.id,
+        voice.language === 'en' ? 'Hello, this is a voice preview.' : '你好，这是一段语音试听。',
+        voice.speed
+      );
+      if (audioRef.current) {
+        audioRef.current.pause();
+      }
+      const audio = new Audio(audioUrl);
+      audio.onended = () => setPlayingVoiceId(null);
+      audio.onerror = () => {
+        setPlayingVoiceId(null);
+        alert('试听失败，请检查 SiliconFlow 配置。');
+      };
+      audioRef.current = audio;
+      setPlayingVoiceId(voice.id);
+      await audio.play();
+    } catch (error: any) {
+      alert(error?.message || '试听失败');
+      setPlayingVoiceId(null);
+    } finally {
+      setPlayLoadingId(null);
     }
   };
 
-  const handleAddSuccess = (newVoice: Voice) => {
+  const handleAddSuccess = async (newVoice: Voice) => {
-      setVoices([newVoice, ...voices]);
+      const created = await createVoice(newVoice);
+      setVoices((prev) => [created, ...prev]);
       setIsAddModalOpen(false);
       setIsCloneModalOpen(false);
   };
 
+  const handleUpdateSuccess = async (id: string, data: Voice) => {
+      const updated = await updateVoice(id, data);
+      setVoices((prev) => prev.map((voice) => (voice.id === id ? updated : voice)));
+      setEditingVoice(null);
+  };
+
+  const handleDelete = async (id: string) => {
+      if (!confirm('确认删除这个声音吗？')) return;
+      await deleteVoice(id);
+      setVoices((prev) => prev.filter((voice) => voice.id !== id));
+  };
+
+  const handleSaveVendorCredential = async () => {
+      if (!credentialApiKey.trim()) {
+        alert('请填写 API Key');
+        return;
+      }
+      try {
+        setIsSavingCredential(true);
+        const option = VENDOR_OPTIONS.find((item) => item.key === credentialVendorKey);
+        const saved = await saveVendorCredential(credentialVendorKey, {
+          vendorName: option?.label || credentialVendorKey,
+          apiKey: credentialApiKey.trim(),
+          baseUrl: credentialBaseUrl.trim(),
+        });
+        setVendorCredentials((prev) => ({ ...prev, [saved.vendorKey]: saved }));
+      } catch (error: any) {
+        alert(error?.message || '保存厂商配置失败');
+      } finally {
+        setIsSavingCredential(false);
+      }
+  };
+
   return (
     <div className="space-y-6 animate-in fade-in py-4 pb-10">
       <div className="flex items-center justify-between">
@@ -94,6 +176,7 @@ export const VoiceLibraryPage: React.FC = () => {
             >
                 <option value="all">所有厂商</option>
                 <option value="硅基流动">硅基流动 (SiliconFlow)</option>
+                <option value="SiliconFlow">SiliconFlow</option>
                 <option value="Ali">阿里 (Ali)</option>
                 <option value="Volcano">火山 (Volcano)</option>
                 <option value="Minimax">Minimax</option>
@@ -123,6 +206,34 @@ export const VoiceLibraryPage: React.FC = () => {
          </div>
       </div>
 
+      <div className="grid grid-cols-1 md:grid-cols-4 gap-3 bg-card/50 p-4 rounded-lg border border-white/5 shadow-sm">
+         <select
+            className="flex h-9 w-full rounded-md border-0 bg-white/5 px-3 py-1 text-sm shadow-sm transition-colors focus-visible:outline-none focus-visible:ring-1 focus-visible:ring-primary/50 [&>option]:bg-card text-foreground"
+            value={credentialVendorKey}
+            onChange={(e) => setCredentialVendorKey(e.target.value)}
+         >
+            {VENDOR_OPTIONS.map((item) => (
+              <option key={item.key} value={item.key}>{item.label}</option>
+            ))}
+         </select>
+         <Input
+            type="password"
+            placeholder="Vendor API Key (持久化到后端)"
+            className="border-0 bg-white/5"
+            value={credentialApiKey}
+            onChange={e => setCredentialApiKey(e.target.value)}
+         />
+         <Input
+            placeholder="Base URL (OpenAI compatible, 选填)"
+            className="border-0 bg-white/5"
+            value={credentialBaseUrl}
+            onChange={e => setCredentialBaseUrl(e.target.value)}
+         />
+         <Button onClick={handleSaveVendorCredential} disabled={isSavingCredential}>
+            {isSavingCredential ? '保存中...' : '保存厂商配置'}
+         </Button>
+      </div>
+
       <div className="rounded-md border border-white/5 bg-card/40 backdrop-blur-md overflow-hidden">
         <table className="w-full text-sm">
           <TableHeader>
@@ -132,6 +243,7 @@ export const VoiceLibraryPage: React.FC = () => {
               <TableHead>性别</TableHead>
               <TableHead>语言</TableHead>
               <TableHead className="text-right">试听</TableHead>
+              <TableHead className="text-right">操作</TableHead>
             </TableRow>
           </TableHeader>
           <tbody>
@@ -155,22 +267,31 @@ export const VoiceLibraryPage: React.FC = () => {
                     <Button 
                         variant="ghost" 
                         size="icon" 
-                        onClick={() => handlePlayToggle(voice.id)}
+                        onClick={() => handlePlayToggle(voice)}
+                        disabled={playLoadingId === voice.id}
                         className={playingVoiceId === voice.id ? "text-primary animate-pulse" : ""}
                     >
                         {playingVoiceId === voice.id ? <Pause className="h-4 w-4" /> : <Play className="h-4 w-4" />}
                     </Button>
                 </TableCell>
+                <TableCell className="text-right">
+                    <Button variant="ghost" size="icon" onClick={() => setEditingVoice(voice)}>
+                        <Pencil className="h-4 w-4" />
+                    </Button>
+                    <Button variant="ghost" size="icon" onClick={() => handleDelete(voice.id)} className="text-red-400">
+                        <Trash2 className="h-4 w-4" />
+                    </Button>
+                </TableCell>
               </TableRow>
             ))}
              {!isLoading && filteredVoices.length === 0 && (
                  <TableRow>
-                     <TableCell colSpan={5} className="text-center py-6 text-muted-foreground">暂无声音数据</TableCell>
+                     <TableCell colSpan={6} className="text-center py-6 text-muted-foreground">暂无声音数据</TableCell>
                  </TableRow>
              )}
              {isLoading && (
                  <TableRow>
-                     <TableCell colSpan={5} className="text-center py-6 text-muted-foreground">加载中...</TableCell>
+                     <TableCell colSpan={6} className="text-center py-6 text-muted-foreground">加载中...</TableCell>
                  </TableRow>
              )}
           </tbody>
@@ -183,6 +304,13 @@ export const VoiceLibraryPage: React.FC = () => {
         onSuccess={handleAddSuccess}
       />
 
+      <AddVoiceModal
+        isOpen={!!editingVoice}
+        onClose={() => setEditingVoice(null)}
+        onSuccess={(voice) => handleUpdateSuccess(editingVoice!.id, voice)}
+        initialVoice={editingVoice || undefined}
+      />
+
       <CloneVoiceModal 
         isOpen={isCloneModalOpen} 
         onClose={() => setIsCloneModalOpen(false)} 
@@ -196,15 +324,17 @@ export const VoiceLibraryPage: React.FC = () => {
 const AddVoiceModal: React.FC<{
     isOpen: boolean;
     onClose: () => void;
-    onSuccess: (voice: Voice) => void;
+    onSuccess: (voice: Voice) => Promise<void>;
-}> = ({ isOpen, onClose, onSuccess }) => {
+    initialVoice?: Voice;
+}> = ({ isOpen, onClose, onSuccess, initialVoice }) => {
     const [vendor, setVendor] = useState<'硅基流动' | 'Ali' | 'Volcano' | 'Minimax'>('硅基流动');
     const [name, setName] = useState('');
     
-    const [sfModel, setSfModel] = useState('fishaudio/fish-speech-1.5');
+    const [sfModel, setSfModel] = useState('FunAudioLLM/CosyVoice2-0.5B');
-    const [sfVoiceId, setSfVoiceId] = useState('fishaudio:amy');
+    const [sfVoiceId, setSfVoiceId] = useState('FunAudioLLM/CosyVoice2-0.5B:anna');
     const [sfSpeed, setSfSpeed] = useState(1);
     const [sfGain, setSfGain] = useState(0);
+    const [sfPitch, setSfPitch] = useState(0);
 
     const [model, setModel] = useState('');
     const [voiceKey, setVoiceKey] = useState('');
@@ -214,40 +344,90 @@ const AddVoiceModal: React.FC<{
     
     const [testInput, setTestInput] = useState('你好，正在测试语音合成效果。');
     const [isAuditioning, setIsAuditioning] = useState(false);
+    const [isSaving, setIsSaving] = useState(false);
+    const testAudioRef = useRef<HTMLAudioElement | null>(null);
 
-    const handleAudition = () => {
+    useEffect(() => {
+        if (!initialVoice) return;
+        const nextVendor = initialVoice.vendor === 'SiliconFlow' ? '硅基流动' : initialVoice.vendor;
+        setVendor((nextVendor as any) || '硅基流动');
+        setName(initialVoice.name || '');
+        setGender(initialVoice.gender || 'Female');
+        setLanguage(initialVoice.language || 'zh');
+        setDescription(initialVoice.description || '');
+        setModel(initialVoice.model || '');
+        setVoiceKey(initialVoice.voiceKey || '');
+        setSfModel(initialVoice.model || 'FunAudioLLM/CosyVoice2-0.5B');
+        setSfVoiceId(initialVoice.voiceKey || 'FunAudioLLM/CosyVoice2-0.5B:anna');
+        setSfSpeed(initialVoice.speed ?? 1);
+        setSfGain(initialVoice.gain ?? 0);
+        setSfPitch(initialVoice.pitch ?? 0);
+    }, [initialVoice, isOpen]);
+
+    const handleAudition = async () => {
         if (!testInput.trim()) return;
-        setIsAuditioning(true);
+        if (!initialVoice?.id) {
-        setTimeout(() => setIsAuditioning(false), 2000);
+            alert('请先创建声音，再进行试听。');
+            return;
+        }
+        try {
+            setIsAuditioning(true);
+            const audioUrl = await previewVoice(initialVoice.id, testInput, sfSpeed);
+            if (testAudioRef.current) {
+                testAudioRef.current.pause();
+            }
+            const audio = new Audio(audioUrl);
+            testAudioRef.current = audio;
+            await audio.play();
+        } catch (error: any) {
+            alert(error?.message || '试听失败');
+        } finally {
+            setIsAuditioning(false);
+        }
     };
 
-    const handleSubmit = () => {
+    const handleSubmit = async () => {
         if (!name) { alert("请填写声音显示名称"); return; }
-        
+
-        let newVoice: Voice = {
+        const newVoice: Voice = {
-            id: `${vendor === '硅基流动' ? 'sf' : 'gen'}-${Date.now()}`,
+            id: initialVoice?.id || `${vendor === '硅基流动' ? 'sf' : 'gen'}-${Date.now()}`,
             name: name,
             vendor: vendor,
             gender: gender,
             language: language,
-            description: description || (vendor === '硅基流动' ? `Model: ${sfModel}` : `Model: ${model}`)
+            description: description || (vendor === '硅基流动' ? `Model: ${sfModel}` : `Model: ${model}`),
+            model: vendor === '硅基流动' ? sfModel : model,
+            voiceKey: vendor === '硅基流动' ? sfVoiceId : voiceKey,
+            speed: sfSpeed,
+            gain: sfGain,
+            pitch: sfPitch,
         };
-
+        try {
-        onSuccess(newVoice);
+            setIsSaving(true);
-        setName('');
+            await onSuccess(newVoice);
-        setVendor('硅基流动');
+            setName('');
-        setDescription('');
+            setVendor('硅基流动');
+            setDescription('');
+            setModel('');
+            setVoiceKey('');
+        } catch (error: any) {
+            alert(error?.message || '保存失败');
+        } finally {
+            setIsSaving(false);
+        }
     };
 
     return (
         <Dialog 
             isOpen={isOpen} 
             onClose={onClose} 
-            title="添加声音"
+            title={initialVoice ? "编辑声音" : "添加声音"}
             footer={
                 <>
                     <Button variant="ghost" onClick={onClose}>取消</Button>
-                    <Button onClick={handleSubmit} className="bg-primary hover:bg-primary/90">确认添加</Button>
+                    <Button onClick={handleSubmit} className="bg-primary hover:bg-primary/90" disabled={isSaving}>
+                        {isSaving ? '保存中...' : (initialVoice ? '保存修改' : '确认添加')}
+                    </Button>
                 </>
             }
         >
@@ -286,9 +466,9 @@ const AddVoiceModal: React.FC<{
                                     value={sfModel}
                                     onChange={e => setSfModel(e.target.value)}
                                 >
+                                    <option value="FunAudioLLM/CosyVoice2-0.5B">FunAudioLLM/CosyVoice2-0.5B</option>
                                     <option value="fishaudio/fish-speech-1.5">fishaudio/fish-speech-1.5</option>
                                     <option value="fishaudio/fish-speech-1.4">fishaudio/fish-speech-1.4</option>
-                                    <option value="ByteDance/SA-Speech">ByteDance/SA-Speech</option>
                                 </select>
                             </div>
                             <div className="space-y-1.5">
@@ -312,6 +492,13 @@ const AddVoiceModal: React.FC<{
                                     <span className="text-[10px] font-mono text-primary bg-primary/10 px-1.5 py-0.5 rounded">{sfGain}dB</span>
                                 </div>
                             </div>
+                            <div className="space-y-1.5">
+                                <label className="text-[10px] font-black text-muted-foreground uppercase tracking-widest block">音调 (Pitch)</label>
+                                <div className="flex items-center space-x-2">
+                                    <input type="range" min="-12" max="12" step="1" value={sfPitch} onChange={e => setSfPitch(parseInt(e.target.value))} className="flex-1 accent-primary" />
+                                    <span className="text-[10px] font-mono text-primary bg-primary/10 px-1.5 py-0.5 rounded">{sfPitch}</span>
+                                </div>
+                            </div>
                         </div>
                     </div>
                 ) : (
@@ -382,7 +569,7 @@ const AddVoiceModal: React.FC<{
                             variant="primary" 
                             size="sm" 
                             onClick={handleAudition} 
-                            disabled={isAuditioning}
+                            disabled={isAuditioning || !initialVoice}
                             className="shrink-0 h-9"
                         >
                             {isAuditioning ? <Pause className="h-3.5 w-3.5 animate-pulse" /> : <Play className="h-3.5 w-3.5" />}
@@ -397,7 +584,7 @@ const AddVoiceModal: React.FC<{
 const CloneVoiceModal: React.FC<{ 
     isOpen: boolean; 
     onClose: () => void; 
-    onSuccess: (voice: Voice) => void 
+    onSuccess: (voice: Voice) => Promise<void> 
 }> = ({ isOpen, onClose, onSuccess }) => {
     const [name, setName] = useState('');
     const [description, setDescription] = useState('');
@@ -410,7 +597,7 @@ const CloneVoiceModal: React.FC<{
         }
     };
 
-    const handleSubmit = () => {
+    const handleSubmit = async () => {
         if (!name || !file) {
             alert("请填写名称并上传音频文件");
             return;
@@ -425,7 +612,7 @@ const CloneVoiceModal: React.FC<{
             description: description || 'User cloned voice'
         };
 
-        onSuccess(newVoice);
+        await onSuccess(newVoice);
         setName('');
         setDescription('');
         setFile(null);

web/services/apiClient.ts

@@ -1,4 +1,4 @@
-const DEFAULT_API_BASE_URL = 'http://localhost:8100/api';
+const DEFAULT_API_BASE_URL = 'http://127.0.0.1:8100/api';
 
 const trimTrailingSlash = (value: string): string => value.replace(/\/+$/, '');
 

web/services/backendApi.ts

@@ -1,4 +1,4 @@
-import { Assistant, CallLog, InteractionDetail, KnowledgeBase, KnowledgeDocument, Voice, Workflow, WorkflowEdge, WorkflowNode } from '../types';
+import { Assistant, CallLog, InteractionDetail, KnowledgeBase, KnowledgeDocument, VendorCredential, Voice, Workflow, WorkflowEdge, WorkflowNode } from '../types';
 import { apiRequest } from './apiClient';
 
 type AnyRecord = Record<string, any>;
@@ -46,10 +46,27 @@ const mapAssistant = (raw: AnyRecord): Assistant => ({
 const mapVoice = (raw: AnyRecord): Voice => ({
   id: String(readField(raw, ['id'], '')),
   name: readField(raw, ['name'], ''),
-  vendor: readField(raw, ['vendor'], ''),
+  vendor: ((): string => {
+    const vendor = String(readField(raw, ['vendor'], ''));
+    return vendor.toLowerCase() === 'siliconflow' ? '硅基流动' : vendor;
+  })(),
   gender: readField(raw, ['gender'], ''),
   language: readField(raw, ['language'], ''),
   description: readField(raw, ['description'], ''),
+  model: readField(raw, ['model'], ''),
+  voiceKey: readField(raw, ['voiceKey', 'voice_key'], ''),
+  speed: Number(readField(raw, ['speed'], 1)),
+  gain: Number(readField(raw, ['gain'], 0)),
+  pitch: Number(readField(raw, ['pitch'], 0)),
+  enabled: Boolean(readField(raw, ['enabled'], true)),
+  isSystem: Boolean(readField(raw, ['isSystem', 'is_system'], false)),
+});
+
+const mapVendorCredential = (raw: AnyRecord): VendorCredential => ({
+  vendorKey: String(readField(raw, ['vendorKey', 'vendor_key'], '')),
+  vendorName: readField(raw, ['vendorName', 'vendor_name'], ''),
+  apiKey: readField(raw, ['apiKey', 'api_key'], ''),
+  baseUrl: readField(raw, ['baseUrl', 'base_url'], ''),
 });
 
 const mapWorkflowNode = (raw: AnyRecord): WorkflowNode => ({
@@ -173,13 +190,85 @@ export const deleteAssistant = async (id: string): Promise<void> => {
 };
 
 export const fetchVoices = async (): Promise<Voice[]> => {
-  const response = await apiRequest<AnyRecord[]>('/voices');
+  const response = await apiRequest<{ list?: AnyRecord[] } | AnyRecord[]>('/voices');
-  return response.map((item) => mapVoice(item));
+  const list = Array.isArray(response) ? response : (response.list || []);
+  return list.map((item) => mapVoice(item));
+};
+
+export const createVoice = async (data: Partial<Voice>): Promise<Voice> => {
+  const payload = {
+    id: data.id || undefined,
+    name: data.name || 'New Voice',
+    vendor: data.vendor === '硅基流动' ? 'SiliconFlow' : (data.vendor || 'SiliconFlow'),
+    gender: data.gender || 'Female',
+    language: data.language || 'zh',
+    description: data.description || '',
+    model: data.model || undefined,
+    voice_key: data.voiceKey || undefined,
+    speed: data.speed ?? 1,
+    gain: data.gain ?? 0,
+    pitch: data.pitch ?? 0,
+    enabled: data.enabled ?? true,
+  };
+  const response = await apiRequest<AnyRecord>('/voices', { method: 'POST', body: payload });
+  return mapVoice(response);
+};
+
+export const updateVoice = async (id: string, data: Partial<Voice>): Promise<Voice> => {
+  const payload = {
+    name: data.name,
+    vendor: data.vendor === '硅基流动' ? 'SiliconFlow' : data.vendor,
+    gender: data.gender,
+    language: data.language,
+    description: data.description,
+    model: data.model,
+    voice_key: data.voiceKey,
+    speed: data.speed,
+    gain: data.gain,
+    pitch: data.pitch,
+    enabled: data.enabled,
+  };
+  const response = await apiRequest<AnyRecord>(`/voices/${id}`, { method: 'PUT', body: payload });
+  return mapVoice(response);
+};
+
+export const deleteVoice = async (id: string): Promise<void> => {
+  await apiRequest(`/voices/${id}`, { method: 'DELETE' });
+};
+
+export const previewVoice = async (id: string, text: string, speed?: number, apiKey?: string): Promise<string> => {
+  const response = await apiRequest<{ success: boolean; audio_url?: string; error?: string }>(`/voices/${id}/preview`, {
+    method: 'POST',
+    body: { text, speed, api_key: apiKey },
+  });
+  if (!response.success || !response.audio_url) {
+    throw new Error(response.error || 'Preview failed');
+  }
+  return response.audio_url;
+};
+
+export const fetchVendorCredentials = async (): Promise<VendorCredential[]> => {
+  const response = await apiRequest<{ list?: AnyRecord[] }>('/voices/vendors/credentials');
+  const list = response.list || [];
+  return list.map((item) => mapVendorCredential(item));
+};
+
+export const saveVendorCredential = async (vendorKey: string, data: { vendorName: string; apiKey: string; baseUrl?: string }): Promise<VendorCredential> => {
+  const response = await apiRequest<AnyRecord>(`/voices/vendors/credentials/${vendorKey}`, {
+    method: 'PUT',
+    body: {
+      vendor_name: data.vendorName,
+      api_key: data.apiKey,
+      base_url: data.baseUrl || undefined,
+    },
+  });
+  return mapVendorCredential(response);
 };
 
 export const fetchWorkflows = async (): Promise<Workflow[]> => {
-  const response = await apiRequest<AnyRecord[]>('/workflows');
+  const response = await apiRequest<{ list?: AnyRecord[] } | AnyRecord[]>('/workflows');
-  return response.map((item) => mapWorkflow(item));
+  const list = Array.isArray(response) ? response : (response.list || []);
+  return list.map((item) => mapWorkflow(item));
 };
 
 export const fetchWorkflowById = async (id: string): Promise<Workflow> => {
@@ -238,12 +327,12 @@ export const deleteKnowledgeBase = async (kbId: string): Promise<void> => {
 };
 
 export const uploadKnowledgeDocument = async (kbId: string, file: File): Promise<void> => {
-  const params = new URLSearchParams({
+  const payload = {
     name: file.name,
     size: `${(file.size / 1024).toFixed(1)} KB`,
-    file_type: file.type || 'txt',
+    fileType: file.type || 'txt',
-  });
+  };
-  await apiRequest(`/knowledge/bases/${kbId}/documents?${params.toString()}`, { method: 'POST' });
+  await apiRequest(`/knowledge/bases/${kbId}/documents`, { method: 'POST', body: payload });
 };
 
 export const deleteKnowledgeDocument = async (kbId: string, docId: string): Promise<void> => {

web/types.ts

@@ -28,6 +28,20 @@ export interface Voice {
   gender: string;
   language: string;
   description: string;
+  model?: string;
+  voiceKey?: string;
+  speed?: number;
+  gain?: number;
+  pitch?: number;
+  enabled?: boolean;
+  isSystem?: boolean;
+}
+
+export interface VendorCredential {
+  vendorKey: string;
+  vendorName: string;
+  apiKey: string;
+  baseUrl?: string;
 }
 
 export interface KnowledgeBase {