feat: support MiniMax audio provider

2026-06-03 22:50:01 +08:00 · 2026-05-25 11:42:57 +08:00
parent 9af61c4744
commit 766d2699ea
3 changed files with 221 additions and 3 deletions
--- a/app/agent/llm/capability.py
+++ b/app/agent/llm/capability.py
@@ -9,11 +9,12 @@ import subprocess
 from abc import ABC
 from io import BytesIO
 from pathlib import Path
-from typing import Dict, Optional
+from typing import Any, Dict, Optional
 from uuid import uuid4

 from app.core.config import settings
 from app.log import logger
+from app.utils.http import RequestUtils


 class AgentCapabilityProvider(ABC):
@@ -411,6 +412,160 @@ class MiMoAudioProvider(OpenAIChatAudioProvider):
        return model


+class MiniMaxAudioProvider(OpenAIChatAudioProvider):
+    """MiniMax 音频 provider，语音合成使用官方 T2A HTTP 接口。"""
+
+    name = "minimax"
+    DISPLAY_NAME = "MiniMax"
+    DEFAULT_BASE_URL = "https://api.minimaxi.com/v1"
+    DEFAULT_STT_MODEL = "MiniMax-M2.7"
+    DEFAULT_TTS_MODEL = "speech-2.8-turbo"
+    DEFAULT_VOICE = "Chinese (Mandarin)_Lyrical_Voice"
+    AUDIO_INPUT_DATA_URL = True
+    SUPPORTED_TTS_MODELS = frozenset(
+        {
+            "speech-2.8-hd",
+            "speech-2.8-turbo",
+            "speech-2.6-hd",
+            "speech-2.6-turbo",
+            "speech-02-hd",
+            "speech-02-turbo",
+            "speech-01-hd",
+            "speech-01-turbo",
+        }
+    )
+
+    def _build_client(self, api_key: str, base_url: Optional[str]):
+        """构建 MiniMax OpenAI 兼容客户端，兼容用户误填 Anthropic 端点的情况。"""
+        from openai import OpenAI
+
+        return OpenAI(
+            api_key=api_key,
+            base_url=self._normalize_api_base_url(base_url),
+            max_retries=3,
+        )
+
+    @classmethod
+    def _normalize_api_base_url(cls, base_url: Optional[str]) -> str:
+        """归一化 MiniMax API 基础 URL，确保后续可以拼接 OpenAI/T2A 路径。"""
+        normalized = (base_url or cls.DEFAULT_BASE_URL).strip().rstrip("/")
+        if normalized.endswith("/t2a_v2"):
+            normalized = normalized[: -len("/t2a_v2")]
+        for suffix in ("/anthropic/v1", "/openai/v1"):
+            if normalized.endswith(suffix):
+                return normalized[: -len(suffix)] + "/v1"
+        if not normalized.endswith("/v1"):
+            normalized = f"{normalized}/v1"
+        return normalized
+
+    @classmethod
+    def _build_t2a_url(cls, base_url: Optional[str]) -> str:
+        """生成 MiniMax 同步 T2A 接口地址。"""
+        return f"{cls._normalize_api_base_url(base_url)}/t2a_v2"
+
+    def _normalize_stt_model(self) -> str:
+        """将非 MiniMax 的默认转写模型名兜底为 MiniMax 对话模型。"""
+        model = (settings.AUDIO_INPUT_MODEL or "").strip()
+        if not model or model.lower().startswith(("gpt-", "mimo-")):
+            return self.DEFAULT_STT_MODEL
+        return model
+
+    def _normalize_tts_model(self) -> str:
+        """将非 MiniMax 语音模型兜底为官方 T2A 模型。"""
+        model = (settings.AUDIO_OUTPUT_MODEL or "").strip().lower()
+        if model in self.SUPPORTED_TTS_MODELS:
+            return model
+        return self.DEFAULT_TTS_MODEL
+
+    def _normalize_voice_id(self) -> str:
+        """将其他 provider 的默认音色兜底为 MiniMax 中文系统音色。"""
+        voice_id = (settings.AUDIO_OUTPUT_VOICE or "").strip()
+        if not voice_id or voice_id in {"alloy", "mimo_default"}:
+            return self.DEFAULT_VOICE
+        return voice_id
+
+    @staticmethod
+    def _decode_audio_payload(audio_data: str) -> bytes:
+        """解析 MiniMax T2A 返回的音频数据，优先按官方 hex 格式处理。"""
+        normalized = "".join((audio_data or "").split())
+        try:
+            return bytes.fromhex(normalized)
+        except ValueError:
+            return base64.b64decode(audio_data)
+
+    @staticmethod
+    def _extract_minimax_error(data: dict[str, Any]) -> Optional[str]:
+        """提取 MiniMax base_resp 错误信息，成功响应返回 None。"""
+        base_resp = data.get("base_resp") or {}
+        status_code = base_resp.get("status_code")
+        if status_code in (None, 0, "0"):
+            return None
+        status_msg = base_resp.get("status_msg") or "unknown error"
+        return f"{status_code}: {status_msg}"
+
+    def synthesize_speech(self, text: str) -> Optional[Path]:
+        """调用 MiniMax T2A HTTP 接口合成语音文件。"""
+        if not text:
+            return None
+
+        try:
+            api_key, base_url = self._output_credentials()
+            if not api_key:
+                raise ValueError("音频输出 provider 未配置 API Key")
+            response = RequestUtils(
+                headers={
+                    "Authorization": f"Bearer {api_key}",
+                    "Content-Type": "application/json",
+                    "Accept": "application/json",
+                },
+                proxies=settings.PROXY or {},
+                timeout=60,
+            ).post_res(
+                url=self._build_t2a_url(base_url),
+                json={
+                    "model": self._normalize_tts_model(),
+                    "text": text,
+                    "stream": False,
+                    "language_boost": "auto",
+                    "output_format": "hex",
+                    "voice_setting": {
+                        "voice_id": self._normalize_voice_id(),
+                        "speed": 1,
+                        "vol": 1,
+                        "pitch": 0,
+                    },
+                    "audio_setting": {
+                        "sample_rate": 32000,
+                        "bitrate": 128000,
+                        "format": "opus",
+                        "channel": 1,
+                    },
+                },
+            )
+            if not response:
+                raise ValueError("MiniMax T2A 请求无响应")
+            if response.status_code >= 400:
+                raise ValueError(f"MiniMax T2A HTTP {response.status_code}")
+
+            result = response.json()
+            minimax_error = self._extract_minimax_error(result)
+            if minimax_error:
+                raise ValueError(f"MiniMax T2A 返回错误: {minimax_error}")
+
+            audio_data = ((result.get("data") or {}).get("audio") or "").strip()
+            if not audio_data:
+                raise ValueError("MiniMax T2A 响应中没有音频数据")
+
+            voice_dir = settings.TEMP_PATH / "voice"
+            voice_dir.mkdir(parents=True, exist_ok=True)
+            output_path = voice_dir / f"{uuid4().hex}.opus"
+            output_path.write_bytes(self._decode_audio_payload(audio_data))
+            return output_path
+        except Exception as err:
+            logger.error(f"音频输出合成失败: provider={self.name}, error={err}")
+            return None
+
+
 class AgentCapabilityManager:
    """Agent 能力统一入口。"""

@@ -420,6 +575,7 @@ class AgentCapabilityManager:
        OpenAIAudioProvider.name: OpenAIAudioProvider(),
        OpenAIChatAudioProvider.name: OpenAIChatAudioProvider(),
        MiMoAudioProvider.name: MiMoAudioProvider(),
+        MiniMaxAudioProvider.name: MiniMaxAudioProvider(),
    }

    @classmethod
--- a/app/core/config.py
+++ b/app/core/config.py
@@ -600,7 +600,7 @@ class ConfigModel(BaseModel):
    # AI智能体自动重试整理失败记录开关
    AI_AGENT_RETRY_TRANSFER: bool = False

-    # 音频输入提供商：openai/openai_chat_audio/mimo
+    # 音频输入提供商：openai/openai_chat_audio/mimo/minimax
    AUDIO_INPUT_PROVIDER: str = "openai"
    # 音频输入 API 密钥
    AUDIO_INPUT_API_KEY: Optional[str] = None
@@ -610,7 +610,7 @@ class ConfigModel(BaseModel):
    AUDIO_INPUT_MODEL: str = "gpt-4o-mini-transcribe"
    # 音频输入识别语言
    AUDIO_INPUT_LANGUAGE: str = "zh"
-    # 音频输出提供商：openai/openai_chat_audio/mimo
+    # 音频输出提供商：openai/openai_chat_audio/mimo/minimax
    AUDIO_OUTPUT_PROVIDER: str = "openai"
    # 音频输出 API 密钥
    AUDIO_OUTPUT_API_KEY: Optional[str] = None
--- a/tests/test_agent_llm_capability.py
+++ b/tests/test_agent_llm_capability.py
@@ -21,6 +21,7 @@ spec.loader.exec_module(capability_module)

 AgentCapabilityManager = capability_module.AgentCapabilityManager
 MiMoAudioProvider = capability_module.MiMoAudioProvider
+MiniMaxAudioProvider = capability_module.MiniMaxAudioProvider
 OpenAIChatAudioProvider = capability_module.OpenAIChatAudioProvider
 OpenAIAudioProvider = capability_module.OpenAIAudioProvider

@@ -32,6 +33,7 @@ class AgentCapabilityManagerTest(unittest.TestCase):
            "openai_chat_audio", AgentCapabilityManager.get_registered_audio_providers()
        )
        self.assertIn("mimo", AgentCapabilityManager.get_registered_audio_providers())
+        self.assertIn("minimax", AgentCapabilityManager.get_registered_audio_providers())

    def test_get_audio_provider_uses_separate_input_and_output_settings(self):
        with patch.object(settings, "AUDIO_INPUT_PROVIDER", "openai"), patch.object(
@@ -230,6 +232,66 @@ class AgentCapabilityManagerTest(unittest.TestCase):
        )
        self.assertIn("只输出转写结果", content[1]["text"])

+    def test_minimax_stt_normalizes_openai_default_model(self):
+        """校验 MiniMax 音频输入会把 OpenAI 默认模型兜底为 MiniMax 模型。"""
+        provider = MiniMaxAudioProvider()
+
+        with patch.object(settings, "AUDIO_INPUT_MODEL", "gpt-4o-mini-transcribe"):
+            self.assertEqual(provider._normalize_stt_model(), "MiniMax-M2.7")
+
+    def test_minimax_tts_uses_t2a_http_payload(self):
+        """校验 MiniMax 音频输出会调用官方 T2A HTTP 接口并写入音频文件。"""
+        provider = MiniMaxAudioProvider()
+        fake_response = SimpleNamespace(
+            status_code=200,
+            json=Mock(
+                return_value={
+                    "data": {"audio": b"opus-bytes".hex(), "status": 2},
+                    "base_resp": {"status_code": 0, "status_msg": "success"},
+                }
+            ),
+        )
+        request_utils = Mock()
+        request_utils.post_res.return_value = fake_response
+
+        with TemporaryDirectory() as temp_dir, patch.object(
+            capability_module, "RequestUtils", return_value=request_utils
+        ) as request_utils_cls, patch.object(
+            capability_module,
+            "settings",
+            SimpleNamespace(
+                TEMP_PATH=Path(temp_dir),
+                PROXY={},
+                AUDIO_OUTPUT_MODEL="gpt-4o-mini-tts",
+                AUDIO_OUTPUT_VOICE="alloy",
+                AUDIO_OUTPUT_API_KEY="sk-test",
+                AUDIO_OUTPUT_BASE_URL="https://api.minimaxi.com/anthropic/v1",
+            ),
+        ):
+            output_path = provider.synthesize_speech("你好")
+            output_bytes = output_path.read_bytes() if output_path else None
+
+        self.assertIsNotNone(output_path)
+        self.assertEqual(output_bytes, b"opus-bytes")
+        request_utils_cls.assert_called_once()
+        request = request_utils.post_res.call_args.kwargs
+        self.assertEqual(request["url"], "https://api.minimaxi.com/v1/t2a_v2")
+        self.assertEqual(request["json"]["model"], "speech-2.8-turbo")
+        self.assertEqual(
+            request["json"]["voice_setting"]["voice_id"],
+            "Chinese (Mandarin)_Lyrical_Voice",
+        )
+        self.assertEqual(request["json"]["audio_setting"]["format"], "opus")
+
+    def test_minimax_tts_accepts_base64_audio_payload(self):
+        """校验 MiniMax 音频解析兼容部分代理返回的 base64 音频数据。"""
+        provider = MiniMaxAudioProvider()
+
+        self.assertEqual(
+            provider._decode_audio_payload(b64encode(b"opus-bytes").decode("utf-8")),
+            b"opus-bytes",
+        )
+

 if __name__ == "__main__":
    unittest.main()