feat: add ElevenLabs TTS API provider

2026-07-01 01:10:21 +08:00 · 2026-06-10 16:47:14 +08:00
parent 56d2b3fb55
commit 0b22349363
6 changed files with 268 additions and 0 deletions
--- a/astrbot/core/config/default.py
+++ b/astrbot/core/config/default.py
@@ -1807,6 +1807,25 @@ CONFIG_METADATA_2 = {
                        "gemini_tts_voice_name": "Leda",
                        "proxy": "",
                    },
+                    "ElevenLabs TTS(API)": {
+                        "hint": "API Key 从 https://elevenlabs.io/app/settings/api-keys 获取。Voice ID 可在 https://elevenlabs.io/app/voice-library 浏览选择。",
+                        "id": "elevenlabs_tts",
+                        "type": "elevenlabs_tts_api",
+                        "provider": "elevenlabs",
+                        "provider_type": "text_to_speech",
+                        "enable": False,
+                        "api_key": "",
+                        "api_base": "https://api.elevenlabs.io/v1",
+                        "model": "eleven_multilingual_v2",
+                        "elevenlabs-tts-voice-id": "JBFqnCBsd6RMkjVDRZzb",
+                        "elevenlabs-tts-output-format": "mp3_44100_128",
+                        "elevenlabs-tts-stability": "",
+                        "elevenlabs-tts-similarity-boost": "",
+                        "elevenlabs-tts-style": "",
+                        "elevenlabs-tts-use-speaker-boost": True,
+                        "timeout": "20",
+                        "proxy": "",
+                    },
                    "OpenAI Embedding": {
                        "id": "openai_embedding",
                        "type": "openai_embedding",
--- a/astrbot/core/provider/manager.py
+++ b/astrbot/core/provider/manager.py
@@ -467,6 +467,10 @@ class ProviderManager:
                from .sources.gemini_tts_source import (
                    ProviderGeminiTTSAPI as ProviderGeminiTTSAPI,
                )
+            case "elevenlabs_tts_api":
+                from .sources.elevenlabs_tts_source import (
+                    ProviderElevenLabsTTSAPI as ProviderElevenLabsTTSAPI,
+                )
            case "openai_embedding":
                from .sources.openai_embedding_source import (
                    OpenAIEmbeddingProvider as OpenAIEmbeddingProvider,
--- a/astrbot/core/provider/sources/elevenlabs_tts_source.py
+++ b/astrbot/core/provider/sources/elevenlabs_tts_source.py
@@ -0,0 +1,173 @@
+import uuid
+from pathlib import Path
+
+import httpx
+
+from astrbot import logger
+from astrbot.core.utils.astrbot_path import get_astrbot_temp_path
+
+from ..entities import ProviderType
+from ..provider import TTSProvider
+from ..register import register_provider_adapter
+
+SUPPORTED_CONTAINER_OUTPUT_PREFIXES = ("mp3", "wav", "opus")
+RAW_AUDIO_OUTPUT_PREFIXES = ("pcm", "ulaw", "alaw")
+
+
+def _parse_optional_float(
+    provider_config: dict,
+    cfg_name: str,
+) -> float | None:
+    value = provider_config.get(cfg_name, "")
+    if value in ("", None):
+        return None
+    try:
+        parsed = float(value)
+    except (TypeError, ValueError) as exc:
+        raise ValueError(f"{cfg_name} must be a number between 0 and 1.") from exc
+    if not 0 <= parsed <= 1:
+        raise ValueError(f"{cfg_name} must be between 0 and 1.")
+    return parsed
+
+
+def _parse_bool(provider_config: dict, cfg_name: str) -> bool:
+    value = provider_config[cfg_name]
+    if isinstance(value, bool):
+        return value
+    if isinstance(value, int):
+        return bool(value)
+    if isinstance(value, str):
+        normalized = value.strip().lower()
+        if normalized in {"true", "1", "yes", "y", "on"}:
+            return True
+        if normalized in {"false", "0", "no", "n", "off"}:
+            return False
+    raise ValueError(f"{cfg_name} must be a boolean value.")
+
+
+def _normalize_timeout(value: int | str | None) -> int:
+    if value in ("", None):
+        return 20
+    try:
+        timeout = int(value)
+    except (TypeError, ValueError) as exc:
+        raise ValueError("timeout must be a positive integer.") from exc
+    if timeout <= 0:
+        raise ValueError("timeout must be a positive integer.")
+    return timeout
+
+
+def _validate_output_format(output_format: str) -> None:
+    fmt = output_format.lower()
+    if fmt.startswith(RAW_AUDIO_OUTPUT_PREFIXES):
+        raise ValueError(
+            "ElevenLabs raw audio output formats are not supported by this provider. "
+            "Use an mp3, wav, or opus output format instead."
+        )
+    if not fmt.startswith(SUPPORTED_CONTAINER_OUTPUT_PREFIXES):
+        raise ValueError(
+            "Unsupported ElevenLabs output format. "
+            "Use an mp3, wav, or opus output format."
+        )
+
+
+@register_provider_adapter(
+    "elevenlabs_tts_api",
+    "ElevenLabs TTS API",
+    provider_type=ProviderType.TEXT_TO_SPEECH,
+)
+class ProviderElevenLabsTTSAPI(TTSProvider):
+    def __init__(
+        self,
+        provider_config: dict,
+        provider_settings: dict,
+    ) -> None:
+        super().__init__(provider_config, provider_settings)
+        self.api_key = provider_config.get("api_key", "")
+        self.api_base = provider_config.get(
+            "api_base", "https://api.elevenlabs.io/v1"
+        ).removesuffix("/")
+        self.voice_id = provider_config.get(
+            "elevenlabs-tts-voice-id", "JBFqnCBsd6RMkjVDRZzb"
+        )
+        self.model_id = provider_config.get("model", "eleven_multilingual_v2")
+        self.set_model(self.model_id)
+        self.output_format = provider_config.get(
+            "elevenlabs-tts-output-format", "mp3_44100_128"
+        )
+        _validate_output_format(self.output_format)
+
+        # Only send explicitly configured voice settings so the API can apply defaults.
+        self.voice_settings: dict = {}
+        for key, cfg_name in (
+            ("stability", "elevenlabs-tts-stability"),
+            ("similarity_boost", "elevenlabs-tts-similarity-boost"),
+            ("style", "elevenlabs-tts-style"),
+        ):
+            value = _parse_optional_float(provider_config, cfg_name)
+            if value is not None:
+                self.voice_settings[key] = value
+        if "elevenlabs-tts-use-speaker-boost" in provider_config:
+            self.voice_settings["use_speaker_boost"] = _parse_bool(
+                provider_config,
+                "elevenlabs-tts-use-speaker-boost",
+            )
+
+        timeout = _normalize_timeout(provider_config.get("timeout", 20))
+
+        proxy = provider_config.get("proxy", "")
+        if proxy:
+            logger.info(f"[ElevenLabs TTS] 使用代理: {proxy}")
+        self.client = httpx.AsyncClient(
+            timeout=timeout,
+            proxy=proxy or None,
+            trust_env=False,
+        )
+
+    def _output_extension(self) -> str:
+        """Infer the audio file extension from the configured output format."""
+        fmt = self.output_format.lower()
+        if fmt.startswith("mp3"):
+            return "mp3"
+        if fmt.startswith("opus"):
+            return "opus"
+        if fmt.startswith("wav"):
+            return "wav"
+        return "mp3"
+
+    async def get_audio(self, text: str) -> str:
+        url = f"{self.api_base}/text-to-speech/{self.voice_id}"
+        headers = {
+            "xi-api-key": self.api_key,
+            "Content-Type": "application/json",
+        }
+        payload: dict = {
+            "text": text,
+            "model_id": self.model_name,
+        }
+        if self.voice_settings:
+            payload["voice_settings"] = self.voice_settings
+
+        response = await self.client.post(
+            url,
+            headers=headers,
+            params={"output_format": self.output_format},
+            json=payload,
+        )
+        if response.status_code != 200:
+            error_text = response.text[:1024]
+            raise Exception(
+                f"ElevenLabs TTS API 请求失败: {response.status_code}, {error_text}"
+            )
+
+        temp_dir = Path(get_astrbot_temp_path())
+        temp_dir.mkdir(parents=True, exist_ok=True)
+        path = (
+            temp_dir / f"elevenlabs_tts_api_{uuid.uuid4()}.{self._output_extension()}"
+        )
+        path.write_bytes(response.content)
+        return str(path)
+
+    async def terminate(self):
+        if self.client:
+            await self.client.aclose()
--- a/dashboard/src/i18n/locales/en-US/features/config-metadata.json
+++ b/dashboard/src/i18n/locales/en-US/features/config-metadata.json
@@ -1590,6 +1590,30 @@
        "description": "voice",
        "hint": "OpenAI TTS voice. OpenAI defaults: 'alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'."
      },
+      "elevenlabs-tts-voice-id": {
+        "description": "Voice ID",
+        "hint": "ElevenLabs voice ID. Browse and copy voice IDs at https://elevenlabs.io/app/voice-library. Default 'JBFqnCBsd6RMkjVDRZzb' (George)."
+      },
+      "elevenlabs-tts-output-format": {
+        "description": "Output format",
+        "hint": "Audio output format, e.g. 'mp3_44100_128', 'mp3_22050_32', 'wav_44100', or 'opus_48000_128'. Raw PCM/u-law/a-law formats are not supported. Default 'mp3_44100_128'."
+      },
+      "elevenlabs-tts-stability": {
+        "description": "Stability",
+        "hint": "Voice stability, range [0, 1]. Higher is more consistent, lower is more expressive. Leave empty to use the server default."
+      },
+      "elevenlabs-tts-similarity-boost": {
+        "description": "Similarity boost",
+        "hint": "How closely the output matches the original voice, range [0, 1]. Leave empty to use the server default."
+      },
+      "elevenlabs-tts-style": {
+        "description": "Style exaggeration",
+        "hint": "Style exaggeration of the voice, range [0, 1]. Higher values increase latency. Leave empty to use the server default."
+      },
+      "elevenlabs-tts-use-speaker-boost": {
+        "description": "Speaker boost",
+        "hint": "Boost similarity to the original speaker. May slightly increase latency."
+      },
      "mimo-tts-voice": {
        "description": "Voice",
        "hint": "MiMo TTS voice name. Supported values include 'mimo_default', 'default_en', and 'default_zh'."
--- a/dashboard/src/i18n/locales/ru-RU/features/config-metadata.json
+++ b/dashboard/src/i18n/locales/ru-RU/features/config-metadata.json
@@ -1587,6 +1587,30 @@
                "description": "API Base URL",
                "hint": "Голоса OpenAI TTS: alloy, echo и др."
            },
+            "elevenlabs-tts-voice-id": {
+                "description": "ID голоса",
+                "hint": "ID голоса ElevenLabs. Просмотрите и скопируйте ID на https://elevenlabs.io/app/voice-library. По умолчанию 'JBFqnCBsd6RMkjVDRZzb' (George)."
+            },
+            "elevenlabs-tts-output-format": {
+                "description": "Формат вывода",
+                "hint": "Формат аудио, например 'mp3_44100_128', 'mp3_22050_32', 'wav_44100' или 'opus_48000_128'. Raw PCM/u-law/a-law форматы не поддерживаются. По умолчанию 'mp3_44100_128'."
+            },
+            "elevenlabs-tts-stability": {
+                "description": "Стабильность",
+                "hint": "Стабильность голоса, диапазон [0, 1]. Оставьте пустым для значения по умолчанию."
+            },
+            "elevenlabs-tts-similarity-boost": {
+                "description": "Усиление сходства",
+                "hint": "Насколько вывод соответствует исходному голосу, диапазон [0, 1]. Оставьте пустым для значения по умолчанию."
+            },
+            "elevenlabs-tts-style": {
+                "description": "Выразительность стиля",
+                "hint": "Выразительность стиля голоса, диапазон [0, 1]. Высокие значения увеличивают задержку. Оставьте пустым для значения по умолчанию."
+            },
+            "elevenlabs-tts-use-speaker-boost": {
+                "description": "Усиление диктора",
+                "hint": "Усиливает сходство с исходным диктором. Может немного увеличить задержку."
+            },
            "mimo-tts-voice": {
                "description": "Голос",
                "hint": "Имя голоса MiMo TTS. Поддерживаются значения 'mimo_default', 'default_en' и 'default_zh'."
--- a/dashboard/src/i18n/locales/zh-CN/features/config-metadata.json
+++ b/dashboard/src/i18n/locales/zh-CN/features/config-metadata.json
@@ -1592,6 +1592,30 @@
        "description": "voice",
        "hint": "OpenAI TTS 的声音。OpenAI 默认支持：'alloy', 'echo', 'fable', 'onyx', 'nova', 'shimmer'"
      },
+      "elevenlabs-tts-voice-id": {
+        "description": "音色 ID",
+        "hint": "ElevenLabs 音色 ID。可在 https://elevenlabs.io/app/voice-library 浏览并复制音色 ID。默认 'JBFqnCBsd6RMkjVDRZzb'（George）。"
+      },
+      "elevenlabs-tts-output-format": {
+        "description": "输出格式",
+        "hint": "音频输出格式，例如 'mp3_44100_128'、'mp3_22050_32'、'wav_44100'、'opus_48000_128'。不支持裸 PCM/u-law/a-law 格式。默认 'mp3_44100_128'。"
+      },
+      "elevenlabs-tts-stability": {
+        "description": "稳定性",
+        "hint": "音色稳定性，范围 [0, 1]。值越高越稳定，越低越富有表现力。留空则使用服务端默认值。"
+      },
+      "elevenlabs-tts-similarity-boost": {
+        "description": "相似度增强",
+        "hint": "输出与原始音色的接近程度，范围 [0, 1]。留空则使用服务端默认值。"
+      },
+      "elevenlabs-tts-style": {
+        "description": "风格夸张度",
+        "hint": "音色风格的夸张程度，范围 [0, 1]。值越高延迟越大。留空则使用服务端默认值。"
+      },
+      "elevenlabs-tts-use-speaker-boost": {
+        "description": "说话人增强",
+        "hint": "增强与原始说话人的相似度，可能略微增加延迟。"
+      },
      "mimo-tts-voice": {
        "description": "音色",
        "hint": "MiMo TTS 的音色名称。可选值包括 'mimo_default'、'default_en'、'default_zh'。"