Enhance credential management and testing functionality

- Introduce new fields for voice, speed, and language in the AssistantConfig and ProviderCredential models to support TTS and ASR configurations.
- Update the database schema and seeding script to accommodate the new fields, ensuring backward compatibility.
- Implement credential testing endpoints and logic to validate OpenAI-compatible credentials, enhancing user experience and reliability.
- Modify frontend components to include new fields in the credential forms and improve connection testing feedback.
- Refactor related services and API interactions to support the new credential testing feature.

backend/services/config_resolver.py

@@ -58,7 +58,9 @@ async def resolve_runtime_config(
         # 模型/音色:凭证的模型ID优先
         model=(llm.model_id if llm else ""),
         asr=(stt.model_id if stt else ""),
-        voice="",  # 音色无独立列,留空 → service_factory 回退 .env TTS_VOICE
+        voice=(tts.voice if tts else ""),
+        stt_language=(stt.language if stt else ""),
+        tts_speed=(tts.speed if tts else 1.0),
         realtimeModel=(realtime.model_id if realtime else ""),
         # 运行时连接信息(真 key + url):凭证优先,否则 .env 兜底
         llm_api_key=(llm.api_key if llm else config.LLM_API_KEY),

backend/services/credential_tester.py

@@ -0,0 +1,124 @@
+"""OpenAI 兼容模型凭证的最小连通测试。"""
+
+from __future__ import annotations
+
+import io
+import time
+import wave
+
+import httpx
+
+from schemas import CredentialTestRequest, CredentialTestResult
+
+TEST_TIMEOUT_SECONDS = 10.0
+
+
+def _endpoint(base_url: str, path: str) -> str:
+    return f"{base_url.rstrip('/')}/{path.lstrip('/')}"
+
+
+def _silent_wav() -> bytes:
+    buffer = io.BytesIO()
+    with wave.open(buffer, "wb") as wav:
+        wav.setnchannels(1)
+        wav.setsampwidth(2)
+        wav.setframerate(16_000)
+        wav.writeframes(b"\x00\x00" * 1_600)
+    return buffer.getvalue()
+
+
+def _error_detail(response: httpx.Response, api_key: str) -> str:
+    try:
+        body = response.json()
+        detail = (
+            body.get("error", {}).get("message")
+            if isinstance(body, dict) and isinstance(body.get("error"), dict)
+            else body.get("detail") if isinstance(body, dict) else None
+        )
+    except ValueError:
+        detail = None
+    text = str(detail or response.text or response.reason_phrase).strip()
+    return text.replace(api_key, "***")[:300]
+
+
+async def test_openai_credential(
+    config: CredentialTestRequest,
+) -> CredentialTestResult:
+    started = time.perf_counter()
+    headers = {"Authorization": f"Bearer {config.api_key}"}
+
+    try:
+        async with httpx.AsyncClient(timeout=TEST_TIMEOUT_SECONDS) as client:
+            if config.type == "LLM":
+                response = await client.post(
+                    _endpoint(config.api_url, "chat/completions"),
+                    headers=headers,
+                    json={
+                        "model": config.model_id,
+                        "messages": [{"role": "user", "content": "Reply with OK."}],
+                        "max_tokens": 1,
+                        "stream": False,
+                    },
+                )
+            elif config.type == "Embedding":
+                response = await client.post(
+                    _endpoint(config.api_url, "embeddings"),
+                    headers=headers,
+                    json={"model": config.model_id, "input": "ping"},
+                )
+            elif config.type == "ASR":
+                response = await client.post(
+                    _endpoint(config.api_url, "audio/transcriptions"),
+                    headers=headers,
+                    data={
+                        "model": config.model_id,
+                        **({"language": config.language} if config.language else {}),
+                    },
+                    files={"file": ("test.wav", _silent_wav(), "audio/wav")},
+                )
+            elif config.type == "TTS":
+                response = await client.post(
+                    _endpoint(config.api_url, "audio/speech"),
+                    headers=headers,
+                    json={
+                        "model": config.model_id,
+                        "input": "测试",
+                        "voice": config.voice,
+                        "speed": config.speed,
+                    },
+                )
+            else:
+                return CredentialTestResult(
+                    ok=False,
+                    message="暂不支持该资源类型的连通测试",
+                    detail=f"当前仅支持 LLM、Embedding、ASR、TTS，收到 {config.type}",
+                )
+
+        latency_ms = round((time.perf_counter() - started) * 1000)
+        if response.is_success:
+            return CredentialTestResult(
+                ok=True,
+                latency_ms=latency_ms,
+                message="连接成功",
+                detail=f"OpenAI 兼容接口响应正常（HTTP {response.status_code}）",
+            )
+        return CredentialTestResult(
+            ok=False,
+            latency_ms=latency_ms,
+            message=f"连接失败（HTTP {response.status_code}）",
+            detail=_error_detail(response, config.api_key),
+        )
+    except httpx.TimeoutException:
+        return CredentialTestResult(
+            ok=False,
+            latency_ms=round((time.perf_counter() - started) * 1000),
+            message="连接超时",
+            detail=f"服务未在 {TEST_TIMEOUT_SECONDS:g} 秒内响应",
+        )
+    except httpx.RequestError as exc:
+        return CredentialTestResult(
+            ok=False,
+            latency_ms=round((time.perf_counter() - started) * 1000),
+            message="无法连接到模型服务",
+            detail=str(exc)[:300],
+        )

backend/services/pipecat/service_factory.py

@@ -11,6 +11,17 @@ from models import AssistantConfig
 from pipecat.services.openai.llm import OpenAILLMService
 from pipecat.services.openai.stt import OpenAISTTService
 from pipecat.services.openai.tts import OpenAITTSService
+from pipecat.transcriptions.language import Language
+
+
+def _language(value: str) -> Language | None:
+    if not value:
+        return None
+    try:
+        return Language(value)
+    except ValueError:
+        logger.warning(f"忽略不支持的 ASR language: {value}")
+        return None
 
 
 def create_stt(cfg: AssistantConfig):
@@ -22,6 +33,7 @@ def create_stt(cfg: AssistantConfig):
         api_key=cfg.stt_api_key or config.STT_API_KEY,
         base_url=cfg.stt_base_url or config.STT_BASE_URL,
         model=cfg.asr or config.STT_MODEL,
+        language=_language(cfg.stt_language),
     )
 
 
@@ -41,6 +53,7 @@ def create_tts(cfg: AssistantConfig):
         base_url=cfg.tts_base_url or config.TTS_BASE_URL,
         model=config.TTS_MODEL,
         voice=cfg.voice or config.TTS_VOICE,
+        speed=cfg.tts_speed,
     )