feat: add DashScope custom voice enrollment flow

backend/routes/model_registry.py

@@ -20,6 +20,7 @@ from schemas import (
     ModelResourceUpsert,
 )
 from services.auth import require_admin
+from services.dashscope_voice import VoiceRequest, manage_voice
 from services.interface_catalog import validate_fields
 from services.masking import mask_secrets, merge_secrets
 from services.model_resource_tester import test_model_resource
@@ -377,3 +378,16 @@ async def delete_model_resource(
     await session.delete(row)
     await session.commit()
     return {"ok": True}
+
+
+@router.post("/model-resources/dashscope/voice")
+async def manage_dashscope_voice(request: VoiceRequest, session: AsyncSession = Depends(get_session)):
+    """Create/query a voice using draft settings and server-held credentials."""
+    stored = {}
+    if request.resource_id:
+        row = await session.get(ModelResource, request.resource_id)
+        if not row or row.interface_type != "dashscope-tts":
+            raise HTTPException(404, "百炼 TTS 模型资源不存在")
+        stored = row.secrets or {}
+    key = merge_secrets({"apiKey": request.api_key}, stored).get("apiKey", "")
+    return await manage_voice(request, key)

backend/services/dashscope_voice.py

@@ -0,0 +1,67 @@
+"""DashScope voice enrollment, separate from the real-time synthesis path."""
+import base64
+import re
+from typing import Literal
+from urllib.parse import urlsplit
+
+import httpx
+from fastapi import HTTPException
+from pydantic import BaseModel, Field
+from services.pipecat.dashscope_tts import SUPPORTED_MODELS
+
+
+class VoiceRequest(BaseModel):
+    resource_id: str | None = Field(default=None, alias="resourceId")
+    api_key: str = Field(default="", alias="apiKey", max_length=4096)
+    api_url: str = Field(alias="apiUrl", max_length=2048)
+    model: str
+    action: Literal["create_voice", "query_voice"]
+    prefix: str = Field(default="assistant", pattern=r"^[A-Za-z0-9]{1,10}$")
+    audio: str = Field(default="", max_length=14_000_000)
+    voice_id: str = Field(default="", alias="voiceId", max_length=512)
+
+
+def enrollment_payload(request: VoiceRequest):
+    host = urlsplit(request.api_url).hostname or ""
+    allowed = host in {"dashscope.aliyuncs.com", "dashscope-intl.aliyuncs.com"} or re.fullmatch(
+        r"[a-zA-Z0-9-]+\.(cn-beijing|ap-southeast-1)\.maas\.aliyuncs\.com", host)
+    if not allowed:
+        raise HTTPException(400, "音色复刻仅支持百炼北京或新加坡官方域名")
+    if request.model not in SUPPORTED_MODELS or request.model == "cosyvoice-v1":
+        raise HTTPException(400, "当前模型不支持此声音复刻流程")
+    data = {"action": request.action}
+    if request.action == "create_voice":
+        match = re.fullmatch(r"data:audio/(wav|x-wav|mpeg|mp3|mp4|m4a|x-m4a);base64,(.+)", request.audio)
+        if not match:
+            raise HTTPException(400, "请上传 WAV、MP3 或 M4A 音频")
+        try:
+            raw = base64.b64decode(match[2], validate=True)
+        except ValueError:
+            raise HTTPException(400, "音频编码不合法")
+        if not raw or len(raw) > 10 * 1024 * 1024:
+            raise HTTPException(400, "音频必须非空且不超过 10 MB")
+        data.update(target_model=request.model, prefix=request.prefix, audio=request.audio)
+    else:
+        if not request.voice_id:
+            raise HTTPException(400, "缺少音色 ID")
+        data["voice_id"] = request.voice_id
+    return f"https://{host}/api/v1/services/audio/tts/customization", {"model": "voice-enrollment", "input": data}
+
+
+async def manage_voice(request: VoiceRequest, api_key: str):
+    if not api_key:
+        raise HTTPException(400, "请先填写 API Key")
+    url, payload = enrollment_payload(request)
+    try:
+        async with httpx.AsyncClient(timeout=60) as client:
+            response = await client.post(url, json=payload, headers={"Authorization": f"Bearer {api_key}"})
+        if response.is_error:
+            raise HTTPException(502, f"百炼音色服务请求失败（HTTP {response.status_code}），请检查凭据、模型和录音要求")
+        output = response.json()["output"]
+        if request.action == "create_voice":
+            return {"voiceId": output["voice_id"], "status": "DEPLOYING"}
+        if output.get("target_model") != request.model:
+            raise HTTPException(400, "此音色绑定的模型与当前选择不一致")
+        return {"voiceId": request.voice_id, "status": output["status"]}
+    except (httpx.HTTPError, ValueError, KeyError, TypeError):
+        raise HTTPException(502, "音色服务连接异常或返回无效，请稍后查询；创建结果不确定时不要重复创建")

backend/tests/test_dashscope_voice.py

@@ -0,0 +1,25 @@
+import unittest
+from fastapi import HTTPException
+from services.dashscope_voice import VoiceRequest, enrollment_payload
+
+
+class VoiceEnrollmentTests(unittest.TestCase):
+    def request(self, **changes):
+        return VoiceRequest.model_validate({"apiUrl": "wss://dashscope.aliyuncs.com/api-ws/v1/inference", "model": "cosyvoice-v3-plus", "action": "create_voice", "audio": "data:audio/wav;base64,UklGRg==", **changes})
+
+    def test_creation_and_region(self):
+        url, payload = enrollment_payload(self.request())
+        self.assertEqual(url, "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization")
+        self.assertEqual(payload["input"]["target_model"], "cosyvoice-v3-plus")
+
+    def test_no_credentials_to_arbitrary_host(self):
+        with self.assertRaises(HTTPException):
+            enrollment_payload(self.request(apiUrl="https://evil.example"))
+
+    def test_query_needs_id(self):
+        with self.assertRaises(HTTPException):
+            enrollment_payload(self.request(action="query_voice"))
+
+    def test_invalid_audio(self):
+        with self.assertRaises(HTTPException):
+            enrollment_payload(self.request(audio="data:audio/wav;base64,???"))