dsh-audiogen 0.2.0 → 0.3.2

package/lib/index.js

@@ -90,13 +90,13 @@ function isPreset(channel, id) {
 function isOpenAICompatible(channel, mode) {
 	return isPreset(channel, "openai") || /(^|\/)(v\d+\/)?audio\/speech$/i.test(channel.apiUrl.trim()) || channel.preset === "custom" && mode === "tts";
 }
-function isElevenLabs(channel) {
+function isElevenLabs$1(channel) {
 	return isPreset(channel, "elevenlabs") || /elevenlabs/i.test(channel.apiUrl);
 }
 function isMiniMax$1(channel) {
 	return isPreset(channel, "minimax") || /minimax/i.test(channel.apiUrl);
 }
-function isStability(channel) {
+function isStability$1(channel) {
 	return isPreset(channel, "stability") || /stability\.ai/i.test(channel.apiUrl);
 }
 function endpointBase(url) {
@@ -281,15 +281,150 @@ function minimaxApiBase(base) {
 	const trimmed = endpointBase(base);
 	return /\/v1$/i.test(trimmed) ? trimmed : `${trimmed}/v1`;
 }
+/**
+* Resolve the MiniMax voice_id for a TTS request.
+* Priority: explicit voice param → upstream id (if it is not a model name) →
+* model alias (if it is not a model name). MiniMax speech/music model ids
+* (speech-2.8-hd, music-3.0, …) are never treated as voice ids.
+*/
+function resolveMiniMaxVoice(request) {
+	const explicit = request.voice?.trim();
+	if (explicit !== void 0 && explicit !== "") return explicit;
+	for (const candidate of [request.upstream, request.model]) {
+		const value = typeof candidate === "string" ? candidate.trim() : "";
+		if (value === "") continue;
+		if (/^(speech|music|t2a|tts)[-_]/i.test(value)) continue;
+		return value;
+	}
+}
+/**
+* Build the full MiniMax t2a_v2 body. Every official field is carried
+* through — voice_setting (voice_id/speed/vol/pitch/emotion/text_normalization/
+* latex_read), pronunciation_dict.tone, audio_setting (format/sample_rate/
+* bitrate/channel/force_cbr), subtitle_enable, aigc_watermark, language_boost,
+* voice_modify and timbre_weights — so callers and skills can reference them.
+*/
+function buildMiniMaxTTSBody(request, model, voiceId) {
+	const body = {
+		model,
+		text: request.prompt,
+		stream: false,
+		voice_setting: {
+			voice_id: voiceId,
+			speed: request.speed ?? 1,
+			vol: request.vol ?? 1,
+			pitch: request.pitch ?? 0,
+			...request.emotion !== void 0 && request.emotion.trim() !== "" ? { emotion: request.emotion.trim() } : {},
+			...request.textNormalization !== void 0 ? { text_normalization: request.textNormalization } : {},
+			...request.latexRead !== void 0 ? { latex_read: request.latexRead } : {}
+		},
+		audio_setting: {
+			format: request.format ?? "mp3",
+			sample_rate: request.sampleRate ?? 32e3,
+			bitrate: request.bitrate ?? 128e3,
+			channel: request.audioChannel ?? 1,
+			...request.forceCbr !== void 0 ? { force_cbr: request.forceCbr } : {}
+		}
+	};
+	if (request.pronunciationTone !== void 0 && request.pronunciationTone.length > 0) body.pronunciation_dict = { tone: request.pronunciationTone };
+	if (request.subtitleEnable !== void 0) body.subtitle_enable = request.subtitleEnable;
+	if (request.aigcWatermark !== void 0) body.aigc_watermark = request.aigcWatermark;
+	if (request.languageBoost !== void 0 && request.languageBoost.trim() !== "") body.language_boost = request.languageBoost.trim();
+	if (request.voiceModify !== void 0) {
+		const modify = {};
+		if (request.voiceModify.pitch !== void 0) modify.pitch = request.voiceModify.pitch;
+		if (request.voiceModify.intensity !== void 0) modify.intensity = request.voiceModify.intensity;
+		if (request.voiceModify.timbre !== void 0) modify.timbre = request.voiceModify.timbre;
+		if (request.voiceModify.soundEffects !== void 0 && request.voiceModify.soundEffects.trim() !== "") modify.sound_effects = request.voiceModify.soundEffects.trim();
+		if (Object.keys(modify).length > 0) body.voice_modify = modify;
+	}
+	if (request.timbreWeights !== void 0 && request.timbreWeights.length > 0) body.timbre_weights = request.timbreWeights.filter((item) => typeof item?.voiceId === "string" && item.voiceId.trim() !== "" && typeof item.weight === "number").map((item) => ({
+		voice_id: item.voiceId.trim(),
+		weight: item.weight
+	}));
+	return body;
+}
+/** The MiniMax-specific fields only (model/text/stream excluded) — used as the
+*  new-api `metadata` payload when a gateway serves MiniMax TTS at /v1/audio/speech.
+*  The merge keeps the gateway-sent model/input, and voice_setting.voice_id is
+*  carried explicitly so relays that overwrite it still get the right voice. */
+function buildMiniMaxTTSUpload(request, voiceId) {
+	const upload = buildMiniMaxTTSBody(request, "", voiceId);
+	delete upload.model;
+	delete upload.text;
+	delete upload.stream;
+	return upload;
+}
+/**
+* OpenAI-compatible MiniMax TTS path for New API style gateways that do not
+* route the native /v1/t2a_v2. The full native field set is carried inside
+* `metadata`, which new-api's MiniMax TTS relay merges into t2a_v2 upstream.
+*/
+async function minimaxTTSGateway(channel, request, signal, voiceId) {
+	const endpoint = `${minimaxApiBase(channel.apiUrl)}/audio/speech`;
+	const model = (request.upstream ?? request.model) || "speech-2.8-hd";
+	const metadata = buildMiniMaxTTSUpload(request, voiceId);
+	const body = {
+		model,
+		input: request.prompt,
+		voice: voiceId,
+		response_format: request.format ?? "mp3",
+		...request.speed !== void 0 ? { speed: request.speed } : {},
+		...Object.keys(metadata).length > 0 ? { metadata } : {}
+	};
+	return normalizeAudioResponse(await fetchWithTimeout(endpoint, {
+		method: "POST",
+		redirect: "follow",
+		headers: {
+			authorization: `Bearer ${channel.apiKey.trim()}`,
+			"content-type": "application/json",
+			accept: "application/json, audio/mpeg"
+		},
+		body: JSON.stringify(body),
+		signal
+	}, UPSTREAM_TIMEOUT_MS), {
+		apiKey: channel.apiKey,
+		fallbackMime: "audio/mpeg"
+	});
+}
 async function minimax(channel, request, signal) {
 	const base = minimaxApiBase(channel.apiUrl);
 	const model = (request.upstream ?? request.model) || (request.mode === "music" ? "music-3.0" : "speech-2.8-hd");
-	const voice = request.voice ?? request.model ?? "";
-	let endpoint;
-	let body;
+	if (request.mode === "voice_design") {
+		const endpoint = `${base}/voice_design`;
+		const body = {
+			prompt: request.prompt,
+			preview_text: request.previewText ?? request.voice ?? "你好，这是新设计的音色试听。"
+		};
+		const response = await fetchWithTimeout(endpoint, {
+			method: "POST",
+			redirect: "error",
+			headers: {
+				authorization: `Bearer ${channel.apiKey.trim()}`,
+				"content-type": "application/json",
+				accept: "application/json"
+			},
+			body: JSON.stringify(body),
+			signal
+		}, UPSTREAM_TIMEOUT_MS);
+		if (!response.ok) {
+			const detail = await response.text().catch(() => "");
+			throw new AudioGenError(`MiniMax voice design API error (HTTP ${response.status})${detail === "" ? "" : `: ${detail.slice(0, 300)}`}`, "audio-api-error");
+		}
+		const payload = await response.json();
+		if (payload.base_resp?.status_code !== void 0 && payload.base_resp.status_code !== 0) throw new AudioGenError(payload.base_resp.status_msg ?? `MiniMax returned status ${payload.base_resp.status_code}`, "audio-api-error");
+		const encoded = payload.trial_audio ?? "";
+		if (encoded === "") throw new AudioGenError("MiniMax voice design returned no trial audio", "audio-empty-result");
+		const isHex = /^[0-9a-fA-F]+$/.test(encoded) && encoded.length % 2 === 0;
+		return [{
+			data: new Uint8Array(Buffer.from(encoded, isHex ? "hex" : "base64")),
+			mime: "audio/mpeg",
+			...payload.voice_id === void 0 ? {} : { voiceId: payload.voice_id }
+		}];
+	}
 	if (request.mode === "music") {
-		endpoint = `${base}/music_generation`;
-		body = {
+		const endpoint = `${base}/music_generation`;
+		const body = {
 			model,
 			prompt: request.prompt,
 			...request.duration !== void 0 ? { duration: request.duration } : {},
@@ -299,26 +434,31 @@ async function minimax(channel, request, signal) {
 				bitrate: 256e3
 			}
 		};
-	} else {
-		endpoint = `${base}/t2a_v2`;
-		body = {
-			model,
-			text: request.prompt,
-			stream: false,
-			...voice === "" ? {} : { voice_setting: {
-				voice_id: voice,
-				...request.speed !== void 0 ? { speed: request.speed } : {},
-				vol: 1,
-				pitch: 0
-			} },
-			audio_setting: {
-				format: request.format ?? "mp3",
-				sample_rate: 32e3,
-				bitrate: 128e3
-			}
-		};
+		const response = await fetchWithTimeout(endpoint, {
+			method: "POST",
+			redirect: "error",
+			headers: {
+				authorization: `Bearer ${channel.apiKey.trim()}`,
+				"content-type": "application/json",
+				accept: "application/json, audio/mpeg"
+			},
+			body: JSON.stringify(body),
+			signal
+		}, UPSTREAM_TIMEOUT_MS);
+		if (!response.ok) {
+			const detail = await response.text().catch(() => "");
+			throw new AudioGenError(`MiniMax music API error (HTTP ${response.status})${detail === "" ? "" : `: ${detail.slice(0, 300)}`}`, "audio-api-error");
+		}
+		return normalizeAudioResponse(response, {
+			apiKey: channel.apiKey,
+			fallbackMime: "audio/mpeg"
+		});
 	}
-	return normalizeAudioResponse(await fetchWithTimeout(endpoint, {
+	const voiceId = resolveMiniMaxVoice(request);
+	if (voiceId === void 0) throw new AudioGenError("MiniMax TTS 需要指定音色 voice_id（如 male-qn-qingse、female-shaonv）：请在「音色」字段填写，或把音色加入渠道模型目录（alias 可任意、upstream 填 voice_id），也可点「获取可用模型」拉取账号音色列表。", "voice-required");
+	const endpoint = `${base}/t2a_v2`;
+	const body = buildMiniMaxTTSBody(request, model, voiceId);
+	const response = await fetchWithTimeout(endpoint, {
 		method: "POST",
 		redirect: "error",
 		headers: {
@@ -328,10 +468,19 @@ async function minimax(channel, request, signal) {
 		},
 		body: JSON.stringify(body),
 		signal
-	}, UPSTREAM_TIMEOUT_MS), {
+	}, UPSTREAM_TIMEOUT_MS);
+	if (response.ok) return normalizeAudioResponse(response, {
 		apiKey: channel.apiKey,
 		fallbackMime: "audio/mpeg"
 	});
+	const detail = await response.text().catch(() => "");
+	if (!(response.status === 404 && /invalid url|invalid_request_error/i.test(detail))) throw new AudioGenError(`MiniMax TTS API error (HTTP ${response.status})${detail === "" ? "" : `: ${detail.slice(0, 300)}`}`, "audio-api-error");
+	try {
+		return await minimaxTTSGateway(channel, request, signal, voiceId);
+	} catch (gatewayError) {
+		const detailText = gatewayError instanceof AudioGenError ? gatewayError.message : String(gatewayError);
+		throw new AudioGenError(`MiniMax 渠道「${channel.name}」网关未提供原生 TTS 接口：POST ${endpoint} 返回 HTTP 404（Invalid URL，网关未路由 /v1/t2a_v2）；已回退 OpenAI 兼容 ${minimaxApiBase(channel.apiUrl)}/audio/speech 仍失败：${detailText.slice(0, 300)}。请把渠道 API 地址配置为官方 https://api.minimaxi.com（配合 MiniMax 官方密钥），或确认网关已将 /v1/audio/speech 映射到 MiniMax 音色渠道。`, "audio-api-error");
+	}
 }
 async function stabilityAudio(channel, request, signal) {
 	const base = endpointBase(channel.apiUrl);
@@ -392,71 +541,22 @@ async function generateAudio(channel, request, signal) {
 	if (channel.apiUrl.trim() === "") throw new AudioGenError("channel API URL is not configured", "audio-no-endpoint");
 	if (channel.apiKey.trim() === "") throw new AudioGenError("channel API key is not configured", "audio-no-key");
 	if (request.prompt.trim() === "") throw new AudioGenError("audio prompt/text is required", "audio-empty-prompt");
-	if (isElevenLabs(channel)) return elevenLabs(channel, request, signal);
+	if (request.mode === "voice_design" && !isMiniMax$1(channel)) throw new AudioGenError("音色设计当前仅支持 MiniMax 渠道", "voice-design-unsupported");
+	if (isElevenLabs$1(channel)) return elevenLabs(channel, request, signal);
 	if (isMiniMax$1(channel)) return minimax(channel, request, signal);
-	if (isStability(channel)) return stabilityAudio(channel, request, signal);
+	if (isStability$1(channel)) return stabilityAudio(channel, request, signal);
 	if (isOpenAICompatible(channel, request.mode)) return openAITTS(channel, request, signal);
 	return genericAudio(channel, request, signal);
 }
 //#endregion
 //#region src/audio-presets.ts
 const AUDIO_PRESETS = [
-	{
-		id: "openai-tts",
-		name: "OpenAI · TTS",
-		apiUrl: "https://api.openai.com/v1",
-		hint: "OpenAI 官方语音合成接口（/audio/speech）",
-		models: [
-			{
-				alias: "tts-1",
-				id: "tts-1",
-				category: "tts"
-			},
-			{
-				alias: "tts-1-hd",
-				id: "tts-1-hd",
-				category: "tts"
-			},
-			{
-				alias: "gpt-4o-mini-tts",
-				id: "gpt-4o-mini-tts",
-				category: "tts"
-			}
-		]
-	},
-	{
-		id: "elevenlabs",
-		name: "ElevenLabs",
-		apiUrl: "https://api.elevenlabs.io/v1",
-		hint: "ElevenLabs TTS；模型列表请填写你的 Voice ID（如 Rachel / Adam 等别名）",
-		models: [
-			{
-				alias: "Rachel",
-				id: "21m00Tcm4TlvDq8ikWAM",
-				category: "tts"
-			},
-			{
-				alias: "Adam",
-				id: "pNInz6obpgDQGcFmaJgB",
-				category: "tts"
-			},
-			{
-				alias: "Antoni",
-				id: "ErXwobaYiN019PkySvjV",
-				category: "tts"
-			},
-			{
-				alias: "Bella",
-				id: "EXAVITQu4vr4xnSDxMaL",
-				category: "tts"
-			}
-		]
-	},
 	{
 		id: "minimax",
 		name: "MiniMax",
 		apiUrl: "https://api.minimaxi.com",
-		hint: "MiniMax 音色设计 / TTS / 音乐生成；可使用“获取可用模型”拉取账号音色",
+		site: "https://www.minimaxi.com",
+		hint: "MiniMax 官方音频：音色设计 / TTS / 音乐生成；建议点击「获取可用模型」拉取账号音色与模型",
 		models: [
 			{
 				alias: "speech-2.8-hd",
@@ -515,11 +615,56 @@ const AUDIO_PRESETS = [
 			}
 		]
 	},
+	{
+		id: "elevenlabs",
+		name: "ElevenLabs",
+		apiUrl: "https://api.elevenlabs.io/v1",
+		site: "https://elevenlabsai.cn",
+		hint: "ElevenLabs 语音合成（TTS）；建议点击「获取可用模型」拉取音色与模型",
+		models: [
+			{
+				alias: "Rachel",
+				id: "21m00Tcm4TlvDq8ikWAM",
+				category: "tts"
+			},
+			{
+				alias: "Adam",
+				id: "pNInz6obpgDQGcFmaJgB",
+				category: "tts"
+			},
+			{
+				alias: "Antoni",
+				id: "ErXwobaYiN019PkySvjV",
+				category: "tts"
+			},
+			{
+				alias: "Bella",
+				id: "EXAVITQu4vr4xnSDxMaL",
+				category: "tts"
+			},
+			{
+				alias: "eleven_multilingual_v2",
+				id: "eleven_multilingual_v2",
+				category: "tts"
+			},
+			{
+				alias: "eleven_turbo_v2_5",
+				id: "eleven_turbo_v2_5",
+				category: "tts"
+			},
+			{
+				alias: "eleven_flash_v2_5",
+				id: "eleven_flash_v2_5",
+				category: "tts"
+			}
+		]
+	},
 	{
 		id: "stability-audio",
-		name: "Stability AI · 音频",
+		name: "Stability AI（stable-audio）",
 		apiUrl: "https://api.stability.ai/v2beta/audio",
-		hint: "Stability AI 音乐/音效生成（stable-audio 系列）",
+		site: "https://stability.ai/stable-audio",
+		hint: "Stability AI 音乐 / 音效生成（stable-audio 系列）",
 		models: [{
 			alias: "stable-audio-2.0",
 			id: "stable-audio-2.0",
@@ -529,13 +674,6 @@ const AUDIO_PRESETS = [
 			id: "stable-audio-1.0",
 			category: "music"
 		}]
-	},
-	{
-		id: "custom",
-		name: "自定义渠道",
-		apiUrl: "",
-		hint: "任意兼容接口；支持 OpenAI 兼容 TTS，或返回音频字节 / JSON 的通用 POST",
-		models: []
 	}
 ];
 /** Look up one built-in provider by id. */
@@ -547,17 +685,32 @@ function audioPresetById(id) {
 function isMiniMax(channel) {
 	return channel.preset === "minimax" || /minimax/i.test(channel.apiUrl);
 }
+function isElevenLabs(channel) {
+	return channel.preset === "elevenlabs" || /elevenlabs/i.test(channel.apiUrl);
+}
+function isStability(channel) {
+	return channel.preset === "stability" || /stability\.ai/i.test(channel.apiUrl);
+}
 function baseUrl(url) {
 	return url.trim().replace(/\/+$/, "");
 }
+/** Whether an upstream model id is audio-related at all. */
 function categoryFor(id) {
 	const value = id.toLowerCase();
-	if (/(tts|speech|voice|t2a)/i.test(value)) return "tts";
-	if (/(music|song|cover|lyrics)/i.test(value)) return "music";
+	if (/(tts|speech|voice|t2a|talk|narration)/i.test(value)) return "tts";
+	if (/(music|song|cover|lyrics|audio|melody|beat)/i.test(value)) return "music";
 	if (/(sfx|sound.?effect|effect|foley)/i.test(value)) return "sfx";
 }
+async function fetchJson(url, init) {
+	const response = await fetch(url, init);
+	if (!response.ok) {
+		const text = await response.text().catch(() => "");
+		throw new Error(`HTTP ${response.status}${text === "" ? "" : `: ${text.slice(0, 300)}`}`);
+	}
+	return response.json();
+}
 async function postJson(url, apiKey, body) {
-	const response = await fetch(url, {
+	return fetchJson(url, {
 		method: "POST",
 		headers: {
 			authorization: `Bearer ${apiKey.trim()}`,
@@ -565,18 +718,20 @@ async function postJson(url, apiKey, body) {
 		},
 		body: JSON.stringify(body)
 	});
-	if (!response.ok) {
-		const text = await response.text().catch(() => "");
-		throw new Error(`HTTP ${response.status}${text === "" ? "" : `: ${text.slice(0, 300)}`}`);
-	}
-	return response.json();
 }
 /** Discover available models/voices for a channel. */
 async function discoverAudioModels(channel) {
 	if (channel.apiUrl.trim() === "") throw new Error("API URL is not configured");
 	if (channel.apiKey.trim() === "") throw new Error("API key is not configured");
-	if (isMiniMax(channel)) {
-		const payload = await postJson(`${baseUrl(channel.apiUrl).replace(/\/v1$/i, "")}/v1/get_voice`, channel.apiKey, { voice_type: "all" });
+	if (isMiniMax(channel)) return discoverMiniMax(channel);
+	if (isElevenLabs(channel)) return discoverElevenLabs(channel);
+	if (isStability(channel)) return discoverStability(channel);
+	return discoverOpenAICompatible(channel);
+}
+async function discoverMiniMax(channel) {
+	const url = `${baseUrl(channel.apiUrl).replace(/\/v1$/i, "")}/v1/get_voice`;
+	try {
+		const payload = await postJson(url, channel.apiKey, { voice_type: "all" });
 		if (payload.base_resp?.status_code !== void 0 && payload.base_resp.status_code !== 0) throw new Error(payload.base_resp.status_msg ?? `MiniMax returned status ${payload.base_resp.status_code}`);
 		const models = [];
 		for (const voice of payload.system_voice ?? []) {
@@ -616,18 +771,80 @@ async function discoverAudioModels(channel) {
 		});
 		return {
 			models: dedupe(models),
-			source: "MiniMax get_voice + built-in music catalog"
+			source: "MiniMax get_voice + music 目录"
+		};
+	} catch (error) {
+		const fallback = (audioPresetById("minimax")?.models ?? []).map((model) => ({ ...model }));
+		const message = error instanceof Error ? error.message : String(error);
+		return {
+			models: dedupe(fallback),
+			source: `内置 MiniMax 目录（音色发现失败：${message.slice(0, 160)}）`
 		};
 	}
-	const url = `${baseUrl(channel.apiUrl)}/models`;
-	const response = await fetch(url, { headers: { authorization: `Bearer ${channel.apiKey.trim()}` } });
-	if (!response.ok) throw new Error(`model list request failed (HTTP ${response.status}); please add models manually`);
-	const payload = await response.json();
+}
+async function discoverElevenLabs(channel) {
+	const base = baseUrl(channel.apiUrl);
+	const headers = { "xi-api-key": channel.apiKey.trim() };
+	const failures = [];
+	const models = [];
+	try {
+		const payload = await fetchJson(`${base}/models`, { headers });
+		for (const item of Array.isArray(payload) ? payload : []) {
+			const id = item.model_id?.trim() ?? "";
+			if (id === "") continue;
+			if (item.capabilities?.text_to_speech !== true && item.capabilities?.voice_change !== true) continue;
+			models.push({
+				alias: item.name?.trim() || id,
+				id,
+				category: "tts",
+				...item.description !== void 0 && item.description.trim() !== "" ? { description: item.description.trim() } : {}
+			});
+		}
+	} catch (error) {
+		failures.push(`模型列表：${error instanceof Error ? error.message : String(error)}`);
+	}
+	try {
+		const payload = await fetchJson(`${base}/voices`, { headers });
+		for (const voice of Array.isArray(payload?.voices) ? payload.voices : []) {
+			const id = voice.voice_id?.trim() ?? "";
+			if (id === "") continue;
+			models.push({
+				alias: voice.name?.trim() || id,
+				id,
+				category: "tts",
+				...voice.description !== void 0 && voice.description.trim() !== "" ? { description: voice.description.trim() } : {}
+			});
+		}
+	} catch (error) {
+		failures.push(`音色列表：${error instanceof Error ? error.message : String(error)}`);
+	}
+	if (models.length === 0) {
+		const fallback = (audioPresetById("elevenlabs")?.models ?? []).map((model) => ({ ...model }));
+		const detail = failures.length === 0 ? "" : `（发现失败：${failures.join("；").slice(0, 160)}）`;
+		return {
+			models: dedupe(fallback),
+			source: `内置 ElevenLabs 目录${detail}`
+		};
+	}
+	return {
+		models: dedupe(models),
+		source: "ElevenLabs /models + /voices"
+	};
+}
+async function discoverStability(channel) {
+	return {
+		models: dedupe((audioPresetById("stability-audio")?.models ?? []).map((model) => ({ ...model }))),
+		source: "Stability stable-audio 内置目录"
+	};
+}
+async function discoverOpenAICompatible(channel) {
+	const payload = await fetchJson(`${baseUrl(channel.apiUrl)}/models`, { headers: { authorization: `Bearer ${channel.apiKey.trim()}` } });
 	const models = [];
 	for (const item of payload.data ?? []) {
 		const id = item.id?.trim() ?? "";
 		if (id === "") continue;
-		const category = categoryFor(id) ?? "tts";
+		const category = categoryFor(id);
+		if (category === void 0) continue;
 		models.push({
 			alias: id,
 			id,
@@ -636,7 +853,7 @@ async function discoverAudioModels(channel) {
 	}
 	return {
 		models: dedupe(models),
-		source: "OpenAI-compatible /models"
+		source: "OpenAI-compatible /models（仅音频相关）"
 	};
 }
 function dedupe(models) {
@@ -731,13 +948,15 @@ async function appendHistory(entry) {
 		model: entry.model,
 		prompt: entry.prompt,
 		...entry.voice === void 0 ? {} : { voice: entry.voice },
+		...entry.voiceId === void 0 ? {} : { voiceId: entry.voiceId },
 		...entry.speed === void 0 ? {} : { speed: entry.speed },
 		...entry.duration === void 0 ? {} : { duration: entry.duration },
 		...entry.format === void 0 ? {} : { format: entry.format },
 		audio: entry.audio.map((audio) => ({
 			url: audio.url,
 			mime: audio.mime,
-			...audio.duration === void 0 ? {} : { duration: audio.duration }
+			...audio.duration === void 0 ? {} : { duration: audio.duration },
+			...audio.voiceId === void 0 ? {} : { voiceId: audio.voiceId }
 		})),
 		...entry.channelId === void 0 ? {} : { channelId: entry.channelId },
 		...entry.channel === void 0 ? {} : { channel: entry.channel }
@@ -809,18 +1028,49 @@ function messageOf(error) {
 	return error instanceof Error ? error.message : String(error);
 }
 function parseGenerateRequest(body) {
-	const mode = body.mode === "music" ? "music" : body.mode === "sfx" ? "sfx" : "tts";
+	const mode = body.mode === "music" ? "music" : body.mode === "sfx" ? "sfx" : body.mode === "voice_design" ? "voice_design" : "tts";
 	const prompt = typeof body.prompt === "string" ? body.prompt.trim() : "";
 	if (prompt === "") return void 0;
+	const num = (value) => typeof value === "number" && Number.isFinite(value) ? value : void 0;
+	const str = (value) => typeof value === "string" && value.trim() !== "" ? value.trim() : void 0;
+	const flag = (value) => typeof value === "boolean" ? value : void 0;
+	const tone = Array.isArray(body.pronunciationTone) ? body.pronunciationTone.filter((item) => typeof item === "string" && item.trim() !== "").map((item) => item.trim()) : void 0;
+	const voiceModifyRaw = body.voiceModify;
+	const voiceModify = typeof voiceModifyRaw === "object" && voiceModifyRaw !== null ? {
+		...num(voiceModifyRaw.pitch) !== void 0 ? { pitch: num(voiceModifyRaw.pitch) } : {},
+		...num(voiceModifyRaw.intensity) !== void 0 ? { intensity: num(voiceModifyRaw.intensity) } : {},
+		...num(voiceModifyRaw.timbre) !== void 0 ? { timbre: num(voiceModifyRaw.timbre) } : {},
+		...str(voiceModifyRaw.soundEffects) !== void 0 ? { soundEffects: str(voiceModifyRaw.soundEffects) } : {}
+	} : void 0;
+	const timbreWeights = Array.isArray(body.timbreWeights) ? body.timbreWeights.filter((item) => typeof item === "object" && item !== null && typeof item.voiceId === "string" && typeof item.weight === "number").map((item) => ({
+		voiceId: item.voiceId.trim(),
+		weight: item.weight
+	})).filter((item) => item.voiceId !== "") : void 0;
 	return {
 		mode,
 		model: typeof body.model === "string" ? body.model.trim() : "",
 		prompt,
 		...typeof body.voice === "string" && body.voice.trim() !== "" ? { voice: body.voice.trim() } : {},
-		...typeof body.speed === "number" ? { speed: body.speed } : {},
-		...typeof body.duration === "number" ? { duration: body.duration } : {},
+		...typeof body.previewText === "string" && body.previewText.trim() !== "" ? { previewText: body.previewText.trim() } : {},
+		...num(body.speed) !== void 0 ? { speed: num(body.speed) } : {},
+		...num(body.duration) !== void 0 ? { duration: num(body.duration) } : {},
 		...typeof body.format === "string" && body.format.trim() !== "" ? { format: body.format.trim() } : {},
-		...typeof body.channelId === "string" && body.channelId !== "" ? { channelId: body.channelId } : {}
+		...typeof body.channelId === "string" && body.channelId !== "" ? { channelId: body.channelId } : {},
+		...str(body.emotion) !== void 0 ? { emotion: str(body.emotion) } : {},
+		...num(body.vol) !== void 0 ? { vol: num(body.vol) } : {},
+		...num(body.pitch) !== void 0 ? { pitch: num(body.pitch) } : {},
+		...flag(body.textNormalization) !== void 0 ? { textNormalization: flag(body.textNormalization) } : {},
+		...flag(body.latexRead) !== void 0 ? { latexRead: flag(body.latexRead) } : {},
+		...tone !== void 0 && tone.length > 0 ? { pronunciationTone: tone } : {},
+		...num(body.sampleRate) !== void 0 ? { sampleRate: num(body.sampleRate) } : {},
+		...num(body.bitrate) !== void 0 ? { bitrate: num(body.bitrate) } : {},
+		...num(body.audioChannel) !== void 0 ? { audioChannel: num(body.audioChannel) } : {},
+		...flag(body.forceCbr) !== void 0 ? { forceCbr: flag(body.forceCbr) } : {},
+		...flag(body.subtitleEnable) !== void 0 ? { subtitleEnable: flag(body.subtitleEnable) } : {},
+		...flag(body.aigcWatermark) !== void 0 ? { aigcWatermark: flag(body.aigcWatermark) } : {},
+		...str(body.languageBoost) !== void 0 ? { languageBoost: str(body.languageBoost) } : {},
+		...voiceModify !== void 0 ? { voiceModify } : {},
+		...timbreWeights !== void 0 && timbreWeights.length > 0 ? { timbreWeights } : {}
 	};
 }
 function toView(descriptor) {
@@ -863,6 +1113,21 @@ function resolveChannelRequest(request, view) {
 	const target = explicit ?? defaults;
 	const asked = request.model.trim();
 	if (asked === "") {
+		if (request.mode === "voice_design") {
+			if (target === void 0) return {
+				ok: false,
+				code: "no-channels",
+				message: "尚未配置任何渠道"
+			};
+			return {
+				ok: true,
+				request: {
+					...request,
+					channelId: target.id,
+					channel: target.name
+				}
+			};
+		}
 		const alias = target?.models[0]?.alias ?? "";
 		if (alias === "") return {
 			ok: false,
@@ -946,7 +1211,7 @@ function makeRoutes(deps) {
 				const stored = view.channels.find((candidate) => candidate.id === (typeof body?.channelId === "string" ? body.channelId : void 0)) ?? view.channels.find((candidate) => candidate.id === view.defaultChannelId) ?? view.channels[0];
 				const channel = {
 					id: stored?.id ?? "preview",
-					preset: stored?.preset ?? "",
+					preset: typeof body?.preset === "string" ? body.preset.trim() : stored?.preset ?? "",
 					name: stored?.name ?? "",
 					apiUrl: typeof body?.apiUrl === "string" && body.apiUrl.trim() !== "" ? body.apiUrl.trim() : stored?.apiUrl ?? "",
 					apiKey: typeof body?.apiKey === "string" && body.apiKey.trim() !== "" ? body.apiKey.trim() : stored?.apiKey ?? "",
@@ -1063,7 +1328,8 @@ function makeRoutes(deps) {
 							b64: Buffer.from(output.data).toString("base64"),
 							mime: saved.mime,
 							bytes: saved.bytes,
-							url: `${AUDIO_API.file}/${encodeURIComponent(saved.file)}`
+							url: `${AUDIO_API.file}/${encodeURIComponent(saved.file)}`,
+							...output.voiceId === void 0 ? {} : { voiceId: output.voiceId }
 						});
 					}
 					let history;
@@ -1220,7 +1486,8 @@ const resultSchema = {
 			enum: [
 				"tts",
 				"music",
-				"sfx"
+				"sfx",
+				"voice_design"
 			]
 		},
 		model: {
@@ -1249,7 +1516,8 @@ const resultSchema = {
 					bytes: {
 						type: "integer",
 						required: true
-					}
+					},
+					voiceId: { type: "string" }
 				}
 			}
 		},
@@ -1287,7 +1555,7 @@ function ensureConfigured(config) {
 function registerAgentAudioTools(ctx, resolve) {
 	return ctx.tools.register(defineTool({
 		name: "generate_audio",
-		description: "Generate audio with the configured audio provider. Supports text-to-speech, music generation and sound effects. The tool call waits for the upstream result and returns same-origin audio URLs; pass those URLs to the user for playback or download. If multiple models are configured, first ask the user which one to use or pass model explicitly.",
+		description: "Generate audio with the configured audio provider. Supports text-to-speech, music generation, sound effects and MiniMax voice design. The tool call waits for the upstream result and returns same-origin audio URLs; pass those URLs to the user for playback or download. If multiple models are configured, first ask the user which one to use or pass model explicitly.",
 		parameters: {
 			prompt: {
 				type: "string",
@@ -1299,7 +1567,8 @@ function registerAgentAudioTools(ctx, resolve) {
 				enum: [
 					"tts",
 					"music",
-					"sfx"
+					"sfx",
+					"voice_design"
 				],
 				description: "Generation mode. Defaults to tts."
 			},
@@ -1309,11 +1578,15 @@ function registerAgentAudioTools(ctx, resolve) {
 			},
 			voice: {
 				type: "string",
-				description: "Optional voice id/name for TTS providers."
+				description: "Optional voice id/name for TTS providers. Required for MiniMax TTS (e.g. male-qn-qingse, female-shaonv); fetch the account voices in Settings > Plugins > AI Audio."
+			},
+			preview_text: {
+				type: "string",
+				description: "Optional preview text for voice_design."
 			},
 			speed: {
 				type: "number",
-				description: "Optional speaking rate / speed multiplier where supported."
+				description: "Optional speaking rate / speed multiplier where supported. MiniMax range 0.5-2.0 (default 1)."
 			},
 			duration: {
 				type: "number",
@@ -1322,6 +1595,95 @@ function registerAgentAudioTools(ctx, resolve) {
 			format: {
 				type: "string",
 				description: "Output format such as mp3 or wav."
+			},
+			emotion: {
+				type: "string",
+				description: "MiniMax TTS emotion, e.g. happy/sad/angry/nervous/fearful/bored (voice_setting.emotion)."
+			},
+			vol: {
+				type: "number",
+				description: "MiniMax TTS volume 0-10, default 1 (voice_setting.vol)."
+			},
+			pitch: {
+				type: "integer",
+				description: "MiniMax TTS pitch shift -12..12 semitones, default 0 (voice_setting.pitch)."
+			},
+			text_normalization: {
+				type: "boolean",
+				description: "MiniMax TTS text normalization switch (voice_setting.text_normalization)."
+			},
+			latex_read: {
+				type: "boolean",
+				description: "MiniMax TTS math formula reading switch (voice_setting.latex_read)."
+			},
+			pronunciation_tone: {
+				type: "array",
+				items: { type: "string" },
+				description: "MiniMax TTS pronunciation dictionary tone entries, each \"word/pronunciation\", e.g. [\"处理/(chu3)(li3)\", \"危险/dangerous\"] (pronunciation_dict.tone)."
+			},
+			sample_rate: {
+				type: "integer",
+				description: "MiniMax TTS sample rate: 16000/24000/32000/44100/48000, default 32000 (audio_setting.sample_rate)."
+			},
+			bitrate: {
+				type: "integer",
+				description: "MiniMax TTS bitrate in bps: 64000-320000, default 128000 (audio_setting.bitrate)."
+			},
+			channel: {
+				type: "integer",
+				description: "MiniMax TTS audio channels: 1 or 2, default 1 (audio_setting.channel)."
+			},
+			force_cbr: {
+				type: "boolean",
+				description: "MiniMax TTS force CBR encoding (audio_setting.force_cbr)."
+			},
+			subtitle_enable: {
+				type: "boolean",
+				description: "MiniMax TTS subtitle output switch (subtitle_enable)."
+			},
+			aigc_watermark: {
+				type: "boolean",
+				description: "MiniMax TTS AIGC watermark switch (aigc_watermark)."
+			},
+			language_boost: {
+				type: "string",
+				description: "MiniMax TTS language boost, e.g. 中英混读 (language_boost, model-dependent)."
+			},
+			voice_modify: {
+				type: "object",
+				additionalProperties: false,
+				properties: {
+					pitch: {
+						type: "integer",
+						description: "Pitch shift for voice modification."
+					},
+					intensity: {
+						type: "integer",
+						description: "Intensity for voice modification."
+					},
+					timbre: {
+						type: "integer",
+						description: "Timbre shift for voice modification."
+					},
+					sound_effects: {
+						type: "string",
+						description: "Sound effect for voice modification, e.g. 耳语."
+					}
+				},
+				description: "MiniMax TTS voice modification (voice_modify, supported by speech-2.8+)."
+			},
+			timbre_weights: {
+				type: "array",
+				items: {
+					type: "object",
+					additionalProperties: false,
+					properties: {
+						voice_id: { type: "string" },
+						weight: { type: "integer" }
+					},
+					required: ["voice_id", "weight"]
+				},
+				description: "MiniMax TTS dual-voice blend weights (timbre_weights)."
 			}
 		},
 		output: {
@@ -1333,18 +1695,57 @@ function registerAgentAudioTools(ctx, resolve) {
 		async execute(args, exec) {
 			const config = resolve();
 			ensureConfigured(config);
-			const picked = resolveModel(config, args.model);
+			const mode = args.mode === "music" ? "music" : args.mode === "sfx" ? "sfx" : args.mode === "voice_design" ? "voice_design" : "tts";
+			const picked = mode === "voice_design" ? (() => {
+				const usable = config.channels.filter((channel) => channel.apiUrl.trim() !== "" && channel.apiKey.trim() !== "");
+				const target = usable.find((channel) => channel.id === config.defaultChannelId) ?? usable[0];
+				if (target === void 0) throw new AudioGenError("No usable audio channel is configured for voice design.", "no-channel-available");
+				return {
+					channel: target,
+					alias: "",
+					upstream: ""
+				};
+			})() : resolveModel(config, args.model);
+			const voiceModify = typeof args.voice_modify === "object" && args.voice_modify !== null ? (() => {
+				const raw = args.voice_modify;
+				const out = {};
+				if (typeof raw.pitch === "number") out.pitch = raw.pitch;
+				if (typeof raw.intensity === "number") out.intensity = raw.intensity;
+				if (typeof raw.timbre === "number") out.timbre = raw.timbre;
+				if (typeof raw.sound_effects === "string" && raw.sound_effects.trim() !== "") out.soundEffects = raw.sound_effects.trim();
+				return Object.keys(out).length > 0 ? out : void 0;
+			})() : void 0;
+			const timbreWeights = Array.isArray(args.timbre_weights) ? args.timbre_weights.filter((item) => typeof item === "object" && item !== null && typeof item.voice_id === "string" && typeof item.weight === "number").map((item) => ({
+				voiceId: item.voice_id.trim(),
+				weight: item.weight
+			})).filter((item) => item.voiceId !== "") : void 0;
 			const request = {
-				mode: args.mode === "music" ? "music" : args.mode === "sfx" ? "sfx" : "tts",
+				mode,
 				model: picked.alias,
 				upstream: picked.upstream,
 				channelId: picked.channel.id,
 				channel: picked.channel.name,
 				prompt: args.prompt.trim(),
 				...typeof args.voice === "string" && args.voice.trim() !== "" ? { voice: args.voice.trim() } : {},
+				...typeof args.preview_text === "string" && args.preview_text.trim() !== "" ? { previewText: args.preview_text.trim() } : {},
 				...typeof args.speed === "number" ? { speed: args.speed } : {},
 				...typeof args.duration === "number" ? { duration: args.duration } : {},
-				...typeof args.format === "string" && args.format.trim() !== "" ? { format: args.format.trim() } : {}
+				...typeof args.format === "string" && args.format.trim() !== "" ? { format: args.format.trim() } : {},
+				...typeof args.emotion === "string" && args.emotion.trim() !== "" ? { emotion: args.emotion.trim() } : {},
+				...typeof args.vol === "number" && Number.isFinite(args.vol) ? { vol: args.vol } : {},
+				...typeof args.pitch === "number" && Number.isFinite(args.pitch) ? { pitch: args.pitch } : {},
+				...typeof args.text_normalization === "boolean" ? { textNormalization: args.text_normalization } : {},
+				...typeof args.latex_read === "boolean" ? { latexRead: args.latex_read } : {},
+				...Array.isArray(args.pronunciation_tone) && args.pronunciation_tone.length > 0 ? { pronunciationTone: args.pronunciation_tone.filter((item) => typeof item === "string" && item.trim() !== "").map((item) => item.trim()) } : {},
+				...typeof args.sample_rate === "number" && Number.isFinite(args.sample_rate) ? { sampleRate: args.sample_rate } : {},
+				...typeof args.bitrate === "number" && Number.isFinite(args.bitrate) ? { bitrate: args.bitrate } : {},
+				...typeof args.channel === "number" && Number.isFinite(args.channel) ? { audioChannel: args.channel } : {},
+				...typeof args.force_cbr === "boolean" ? { forceCbr: args.force_cbr } : {},
+				...typeof args.subtitle_enable === "boolean" ? { subtitleEnable: args.subtitle_enable } : {},
+				...typeof args.aigc_watermark === "boolean" ? { aigcWatermark: args.aigc_watermark } : {},
+				...typeof args.language_boost === "string" && args.language_boost.trim() !== "" ? { languageBoost: args.language_boost.trim() } : {},
+				...voiceModify !== void 0 ? { voiceModify } : {},
+				...timbreWeights !== void 0 && timbreWeights.length > 0 ? { timbreWeights } : {}
 			};
 			try {
 				const outputs = await generateAudio(picked.channel, request, exec.signal);
@@ -1355,7 +1756,8 @@ function registerAgentAudioTools(ctx, resolve) {
 						id: saved.id,
 						url: `/api/dsh-audiogen/audio/${encodeURIComponent(saved.file)}`,
 						mime: saved.mime,
-						bytes: saved.bytes
+						bytes: saved.bytes,
+						...output.voiceId === void 0 ? {} : { voiceId: output.voiceId }
 					});
 				}
 				try {
@@ -1374,7 +1776,8 @@ function registerAgentAudioTools(ctx, resolve) {
 							b64: Buffer.from(output.data).toString("base64"),
 							mime: audio[index].mime,
 							bytes: audio[index].bytes,
-							url: audio[index].url
+							url: audio[index].url,
+							...output.voiceId === void 0 ? {} : { voiceId: output.voiceId }
 						})),
 						channelId: picked.channel.id,
 						channel: picked.channel.name