dsh-audiogen 0.2.0 → 0.3.0

package/lib/index.js

@@ -285,6 +285,38 @@ 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 ?? "";
+	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 }
+		}];
+	}
 	let endpoint;
 	let body;
 	if (request.mode === "music") {
@@ -392,6 +424,7 @@ 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 (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(channel)) return elevenLabs(channel, request, signal);
 	if (isMiniMax$1(channel)) return minimax(channel, request, signal);
 	if (isStability(channel)) return stabilityAudio(channel, request, signal);
@@ -731,13 +764,15 @@ 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 }
@@ -809,7 +844,7 @@ 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;
 	return {
@@ -817,6 +852,7 @@ function parseGenerateRequest(body) {
 		model: typeof body.model === "string" ? body.model.trim() : "",
 		prompt,
 		...typeof body.voice === "string" && body.voice.trim() !== "" ? { voice: body.voice.trim() } : {},
+		...typeof body.previewText === "string" && body.previewText.trim() !== "" ? { previewText: body.previewText.trim() } : {},
 		...typeof body.speed === "number" ? { speed: body.speed } : {},
 		...typeof body.duration === "number" ? { duration: body.duration } : {},
 		...typeof body.format === "string" && body.format.trim() !== "" ? { format: body.format.trim() } : {},
@@ -863,6 +899,21 @@ 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,
@@ -1063,7 +1114,8 @@ 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 +1272,8 @@ const resultSchema = {
 			enum: [
 				"tts",
 				"music",
-				"sfx"
+				"sfx",
+				"voice_design"
 			]
 		},
 		model: {
@@ -1249,7 +1302,8 @@ const resultSchema = {
 					bytes: {
 						type: "integer",
 						required: true
-					}
+					},
+					voiceId: { type: "string" }
 				}
 			}
 		},
@@ -1287,7 +1341,7 @@ 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 +1353,8 @@ function registerAgentAudioTools(ctx, resolve) {
 				enum: [
 					"tts",
 					"music",
-					"sfx"
+					"sfx",
+					"voice_design"
 				],
 				description: "Generation mode. Defaults to tts."
 			},
@@ -1311,6 +1366,10 @@ function registerAgentAudioTools(ctx, resolve) {
 				type: "string",
 				description: "Optional voice id/name for TTS providers."
 			},
+			preview_text: {
+				type: "string",
+				description: "Optional preview text for voice_design."
+			},
 			speed: {
 				type: "number",
 				description: "Optional speaking rate / speed multiplier where supported."
@@ -1333,15 +1392,26 @@ 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 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() } : {}
@@ -1355,7 +1425,8 @@ 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 +1445,8 @@ 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

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "dsh-audiogen",
   "description": "AI audio generation plugin for the dsh web GUI: multi-vendor TTS/music/sound-effect channels (OpenAI-compatible, ElevenLabs, MiniMax, Stability AI and custom), per-channel model/voice catalogs, Agent tool and a sidebar AI 音频 panel.",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "type": "module",
   "main": "lib/index.js",
   "exports": {

package/src/agent-audio-tools.ts

@@ -24,6 +24,7 @@ interface AgentAudioRef {
   url: string
   mime: string
   bytes: number
+  voiceId?: string
 }
 
 interface AgentAudioResult {
@@ -43,6 +44,7 @@ const audioRefSchema = {
     url: { type: 'string', required: true },
     mime: { type: 'string', required: true },
     bytes: { type: 'integer', required: true },
+    voiceId: { type: 'string' },
   },
 } as const
 
@@ -52,7 +54,7 @@ const resultSchema = {
   properties: {
     status: { type: 'string', required: true },
     message: { type: 'string', required: true },
-    mode: { type: 'string', required: true, enum: ['tts', 'music', 'sfx'] },
+    mode: { type: 'string', required: true, enum: ['tts', 'music', 'sfx', 'voice_design'] },
     model: { type: 'string', required: true },
     audio: { type: 'array', required: true, items: audioRefSchema },
     error: { type: 'string' },
@@ -98,12 +100,13 @@ function ensureConfigured(config: AgentAudioToolConfig): void {
 export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioToolConfig): () => void {
   const disposer = 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', required: true, description: 'For tts, the text to speak. For music/sfx, a descriptive prompt.' },
-      mode: { type: 'string', enum: ['tts', 'music', 'sfx'], description: 'Generation mode. Defaults to tts.' },
+      mode: { type: 'string', enum: ['tts', 'music', 'sfx', 'voice_design'], description: 'Generation mode. Defaults to tts.' },
       model: { type: 'string', description: 'One of the configured audio models/voices. Defaults to the first configured model.' },
       voice: { type: 'string', description: 'Optional voice id/name for TTS providers.' },
+      preview_text: { type: 'string', description: 'Optional preview text for voice_design.' },
       speed: { type: 'number', description: 'Optional speaking rate / speed multiplier where supported.' },
       duration: { type: 'number', description: 'Requested duration in seconds for music/sfx.' },
       format: { type: 'string', description: 'Output format such as mp3 or wav.' },
@@ -117,15 +120,24 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
     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 === undefined) 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 request: GenerateAudioRequest = {
-        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() } : {}),
@@ -140,6 +152,7 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
             url: `/api/dsh-audiogen/audio/${encodeURIComponent(saved.file)}`,
             mime: saved.mime,
             bytes: saved.bytes,
+            ...(output.voiceId === undefined ? {} : { voiceId: output.voiceId }),
           })
         }
         try {
@@ -159,6 +172,7 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
               mime: audio[index]!.mime,
               bytes: audio[index]!.bytes,
               url: audio[index]!.url,
+              ...(output.voiceId === undefined ? {} : { voiceId: output.voiceId }),
             })),
             channelId: picked.channel.id,
             channel: picked.channel.name,

package/src/audio-engine.ts

@@ -164,7 +164,7 @@ async function fetchWithTimeout(url: string, init: RequestInit, timeoutMs: numbe
 async function normalizeAudioResponse(
   response: Response,
   options: { apiKey: string; fallbackMime?: string },
-): Promise<Array<{ data: Uint8Array; mime: string }>> {
+): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   if (!response.ok) {
     let detail = ''
     try {
@@ -213,7 +213,7 @@ async function normalizeAudioResponse(
   return [{ data: buffer, mime: audioMime(buffer, response.headers.get('content-type') ?? contentType ?? null) }]
 }
 
-async function openAITTS(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string }>> {
+async function openAITTS(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   const base = endpointBase(channel.apiUrl)
   const endpoint = /\/audio\/speech(\?|$)/i.test(base) ? base : `${base}/audio/speech`
   const model = (request.upstream ?? request.model) || 'tts-1'
@@ -239,7 +239,7 @@ async function openAITTS(channel: AudioChannel, request: GenerateAudioRequest, s
   return normalizeAudioResponse(response, { apiKey: channel.apiKey, fallbackMime: 'audio/mpeg' })
 }
 
-async function elevenLabs(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string }>> {
+async function elevenLabs(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   const base = endpointBase(channel.apiUrl)
   const model = (request.upstream ?? request.model) || 'eleven_multilingual_v2'
   const voiceId = (request.voice ?? request.model ?? model).trim()
@@ -274,11 +274,51 @@ function minimaxApiBase(base: string): string {
   return /\/v1$/i.test(trimmed) ? trimmed : `${trimmed}/v1`
 }
 
-async function minimax(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string }>> {
+async function minimax(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   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 ?? ''
 
+  if (request.mode === 'voice_design') {
+    const endpoint = `${base}/voice_design`
+    const body: Record<string, unknown> = {
+      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() as {
+      voice_id?: string
+      trial_audio?: string
+      base_resp?: { status_code?: number; status_msg?: string }
+    }
+    if (payload.base_resp?.status_code !== undefined && 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
+    const data = new Uint8Array(Buffer.from(encoded, isHex ? 'hex' : 'base64'))
+    return [{
+      data,
+      mime: 'audio/mpeg',
+      ...(payload.voice_id === undefined ? {} : { voiceId: payload.voice_id }),
+    }]
+  }
+
   let endpoint: string
   let body: Record<string, unknown>
   if (request.mode === 'music') {
@@ -327,7 +367,7 @@ async function minimax(channel: AudioChannel, request: GenerateAudioRequest, sig
   return normalizeAudioResponse(response, { apiKey: channel.apiKey, fallbackMime: 'audio/mpeg' })
 }
 
-async function stabilityAudio(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string }>> {
+async function stabilityAudio(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   const base = endpointBase(channel.apiUrl)
   const endpoint = /\/generation(\?|$)/i.test(base) ? base : `${base}/generation`
   const model = (request.upstream ?? request.model) || 'stable-audio-2.0'
@@ -351,7 +391,7 @@ async function stabilityAudio(channel: AudioChannel, request: GenerateAudioReque
   return normalizeAudioResponse(response, { apiKey: channel.apiKey, fallbackMime: 'audio/mpeg' })
 }
 
-async function genericAudio(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string }>> {
+async function genericAudio(channel: AudioChannel, request: GenerateAudioRequest, signal?: AbortSignal): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   const base = endpointBase(channel.apiUrl)
   if (request.mode === 'tts' && !/\/generate(\?|$)/i.test(base)) {
     return openAITTS(channel, request, signal)
@@ -388,10 +428,13 @@ export async function generateAudio(
   channel: AudioChannel,
   request: GenerateAudioRequest,
   signal?: AbortSignal,
-): Promise<Array<{ data: Uint8Array; mime: string }>> {
+): Promise<Array<{ data: Uint8Array; mime: string; voiceId?: string }>> {
   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 (request.mode === 'voice_design' && !isMiniMax(channel)) {
+    throw new AudioGenError('音色设计当前仅支持 MiniMax 渠道', 'voice-design-unsupported')
+  }
 
   if (isElevenLabs(channel)) return elevenLabs(channel, request, signal)
   if (isMiniMax(channel)) return minimax(channel, request, signal)

package/src/audio-store.ts

@@ -99,6 +99,7 @@ export async function appendHistory(entry: HistoryEntryInput): Promise<HistoryEn
     model: entry.model,
     prompt: entry.prompt,
     ...(entry.voice === undefined ? {} : { voice: entry.voice }),
+    ...(entry.voiceId === undefined ? {} : { voiceId: entry.voiceId }),
     ...(entry.speed === undefined ? {} : { speed: entry.speed }),
     ...(entry.duration === undefined ? {} : { duration: entry.duration }),
     ...(entry.format === undefined ? {} : { format: entry.format }),
@@ -106,6 +107,7 @@ export async function appendHistory(entry: HistoryEntryInput): Promise<HistoryEn
       url: audio.url,
       mime: audio.mime,
       ...(audio.duration === undefined ? {} : { duration: audio.duration }),
+      ...(audio.voiceId === undefined ? {} : { voiceId: audio.voiceId }),
     })),
     ...(entry.channelId === undefined ? {} : { channelId: entry.channelId }),
     ...(entry.channel === undefined ? {} : { channel: entry.channel }),

package/src/client/AudioGenPanel.tsx

@@ -1,5 +1,8 @@
+
 /**
  * The AI 音频 panel: a compact audio-generation studio.
+ * TTS / music / SFX / voice design are separated; each mode only lists
+ * compatible models and shows its own parameters.
  */
 
 import { useEffect, useMemo, useState } from 'react'
@@ -7,7 +10,7 @@ import type { AudiogenApi } from './api.ts'
 import type { AudiogenScope } from './settings-scope.ts'
 import { audioModelOptions } from './settings-scope.ts'
 import { tt } from './helpers.ts'
-import { GENERATE_API, HISTORY_API, type AudioMode, type GeneratedAudio, type HistoryEntry } from '../protocol.ts'
+import { HISTORY_API, type AudioMode, type GeneratedAudio, type HistoryEntry } from '../protocol.ts'
 import css from './audio-panel.module.css'
 
 function useConfig(scope: AudiogenScope) {
@@ -47,11 +50,12 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
   const channels = config?.channels ?? []
   const connected = enabled && channels.some(channel => {
     const keyHeld = scope.getSecretSetSnapshot(`channelSecrets.${channel.id}`)
-    return channel.apiUrl.trim() !== '' && keyHeld && channel.models.length > 0
+    return channel.apiUrl.trim() !== '' && keyHeld && (channel.models.length > 0 || channel.preset === 'minimax')
   })
 
   const [mode, setMode] = useState<AudioMode>('tts')
   const [prompt, setPrompt] = useState('')
+  const [previewText, setPreviewText] = useState('')
   const [model, setModel] = useState('')
   const [voice, setVoice] = useState('')
   const [speed, setSpeed] = useState('')
@@ -62,9 +66,12 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
   const [outputs, setOutputs] = useState<GeneratedAudio[]>([])
   const { entries, reload, clear } = useHistory()
 
-  const visibleModels = useMemo(() => modelOptions.models
-    .filter(entry => entry.category === undefined || entry.category === 'tts' && mode === 'tts' || entry.category === mode)
-    .map(entry => entry.alias), [modelOptions.models, mode])
+  const visibleModels = useMemo(() => {
+    if (mode === 'voice_design') return []
+    return modelOptions.models
+      .filter(entry => entry.category === undefined || entry.category === 'tts' && mode === 'tts' || entry.category === mode)
+      .map(entry => entry.alias)
+  }, [modelOptions.models, mode])
 
   useEffect(() => {
     if (visibleModels.length > 0 && !visibleModels.includes(model)) {
@@ -84,6 +91,7 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
         mode,
         model: (model || visibleModels[0]) ?? '',
         prompt: prompt.trim(),
+        ...(previewText.trim() !== '' ? { previewText: previewText.trim() } : {}),
         ...(voice.trim() !== '' ? { voice: voice.trim() } : {}),
         ...(speed.trim() !== '' ? { speed: Number(speed) } : {}),
         ...(duration.trim() !== '' ? { duration: Number(duration) } : {}),
@@ -105,9 +113,12 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
   const modeLabel = useMemo(() => {
     if (mode === 'tts') return tt('mode.tts')
     if (mode === 'music') return tt('mode.music')
-    return tt('mode.sfx')
+    if (mode === 'sfx') return tt('mode.sfx')
+    return tt('mode.voiceDesign')
   }, [mode])
 
+  const needModel = mode !== 'voice_design'
+
   return (
     <div className={css.panel}>
       <header className={css.header}>
@@ -117,7 +128,7 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
       <div className={css.layout}>
         <div className={css.form}>
           <div className={css.modeRow}>
-            {(['tts', 'music', 'sfx'] as const).map(item => (
+            {(['tts', 'music', 'sfx', 'voice_design'] as const).map(item => (
               <button
                 key={item}
                 type="button"
@@ -125,50 +136,73 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
                 data-active={mode === item ? 'true' : 'false'}
                 onClick={() => setMode(item)}
               >
-                {item === 'tts' ? tt('mode.tts') : item === 'music' ? tt('mode.music') : tt('mode.sfx')}
+                {item === 'tts' ? tt('mode.tts') : item === 'music' ? tt('mode.music') : item === 'sfx' ? tt('mode.sfx') : tt('mode.voiceDesign')}
               </button>
             ))}
           </div>
+
           <label className={css.label}>
-            <span>{mode === 'tts' ? '文本' : '提示词'}</span>
+            <span>{mode === 'voice_design' ? '音色描述' : mode === 'tts' ? '文本' : '提示词'}</span>
             <textarea className={css.textarea} value={prompt} onChange={event => setPrompt(event.target.value)} placeholder={tt('prompt.placeholder')} />
           </label>
-          <label className={css.label}>
-            <span>{tt('model.label')}</span>
-            <select className={css.select} value={model} onChange={event => setModel(event.target.value)}>
-              {visibleModels.length === 0 ? <option value="">（当前模式暂无可用模型）</option> : null}
-              {visibleModels.map(item => <option key={item} value={item}>{item}</option>)}
-            </select>
-          </label>
+
+          {mode === 'voice_design' ? (
+            <label className={css.label}>
+              <span>试听文本</span>
+              <input className={css.input} value={previewText} onChange={event => setPreviewText(event.target.value)} placeholder="你好，这是新设计的音色试听。" />
+            </label>
+          ) : null}
+
+          {needModel ? (
+            <label className={css.label}>
+              <span>{tt('model.label')}</span>
+              <select className={css.select} value={model} onChange={event => setModel(event.target.value)}>
+                {visibleModels.length === 0 ? <option value="">（当前模式暂无可用模型）</option> : null}
+                {visibleModels.map(item => <option key={item} value={item}>{item}</option>)}
+              </select>
+            </label>
+          ) : null}
+
           {mode === 'tts' ? (
             <label className={css.label}>
               <span>{tt('voice.label')}</span>
               <input className={css.input} value={voice} onChange={event => setVoice(event.target.value)} placeholder="alloy / 自定义音色" />
             </label>
           ) : null}
-          <label className={css.label}>
-            <span>{tt('speed.label')}</span>
-            <input className={css.input} type="number" step="0.1" min="0.5" max="2" value={speed} onChange={event => setSpeed(event.target.value)} placeholder="1.0" />
-          </label>
-          <label className={css.label}>
-            <span>{tt('duration.label')}</span>
-            <input className={css.input} type="number" step="1" min="1" max="120" value={duration} onChange={event => setDuration(event.target.value)} placeholder="30" />
-          </label>
-          <label className={css.label}>
-            <span>{tt('format.label')}</span>
-            <select className={css.select} value={format} onChange={event => setFormat(event.target.value)}>
-              <option value="mp3">mp3</option>
-              <option value="wav">wav</option>
-              <option value="flac">flac</option>
-              <option value="ogg">ogg</option>
-              <option value="pcm">pcm</option>
-            </select>
-          </label>
+
+          {mode === 'tts' ? (
+            <label className={css.label}>
+              <span>{tt('speed.label')}</span>
+              <input className={css.input} type="number" step="0.1" min="0.5" max="2" value={speed} onChange={event => setSpeed(event.target.value)} placeholder="1.0" />
+            </label>
+          ) : null}
+
+          {mode === 'music' || mode === 'sfx' ? (
+            <label className={css.label}>
+              <span>{tt('duration.label')}</span>
+              <input className={css.input} type="number" step="1" min="1" max="120" value={duration} onChange={event => setDuration(event.target.value)} placeholder="30" />
+            </label>
+          ) : null}
+
+          {needModel ? (
+            <label className={css.label}>
+              <span>{tt('format.label')}</span>
+              <select className={css.select} value={format} onChange={event => setFormat(event.target.value)}>
+                <option value="mp3">mp3</option>
+                <option value="wav">wav</option>
+                <option value="flac">flac</option>
+                <option value="ogg">ogg</option>
+                <option value="pcm">pcm</option>
+              </select>
+            </label>
+          ) : null}
+
           {!connected && <p className={css.hint}>{tt('config.missing')}</p>}
-          <button type="button" className={css.generate} disabled={loading || !connected || visibleModels.length === 0} onClick={() => void submit()}>
+          <button type="button" className={css.generate} disabled={loading || !connected || (needModel && visibleModels.length === 0)} onClick={() => void submit()}>
             {loading ? tt('generating') : tt('generate')}
           </button>
         </div>
+
         <div className={css.result}>
           {error !== null ? <p className={css.error}>{error}</p> : null}
           {outputs.length === 0 ? <p className={css.empty}>{tt('result.empty')}</p> : (
@@ -177,6 +211,7 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
               <div className={css.audioList}>
                 {outputs.map((audio, index) => (
                   <div className={css.audioCard} key={audio.id}>
+                    {audio.voiceId !== undefined ? <p className={css.hint}>新音色 ID：{audio.voiceId}</p> : null}
                     <audio className={css.audio} controls preload="metadata" src={dataUrlOf(audio)} />
                     <a className={css.download} href={dataUrlOf(audio)} download={`generated-${index + 1}.${audio.mime.split('/')[1]?.replace('mpeg', 'mp3') ?? 'mp3'}`}>下载</a>
                   </div>
@@ -185,6 +220,7 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
             </>
           )}
         </div>
+
         <aside className={css.history}>
           <div className={css.historyHeader}>
             <strong className={css.historyTitle}>{tt('history.title')}</strong>

package/src/client/locales.ts

@@ -9,7 +9,8 @@ export const zh = {
   'mode.tts': '文本转语音',
   'mode.music': '音乐生成',
   'mode.sfx': '音效生成',
-  'prompt.placeholder': '输入要朗读的文本，或描述想生成的音乐 / 音效…',
+  'mode.voiceDesign': '音色设计',
+  'prompt.placeholder': '输入文本、音乐/音效描述，或音色设计描述…',
   'prompt.required': '请输入文本或提示词',
   'model.label': '模型 / 音色',
   'voice.label': '音色',
@@ -75,7 +76,8 @@ export const en: Record<AudioGenKey, string> = {
   'mode.tts': 'Text to speech',
   'mode.music': 'Music',
   'mode.sfx': 'Sound effects',
-  'prompt.placeholder': 'Text to speak, or a description of the music / sound effect…',
+  'mode.voiceDesign': 'Voice design',
+  'prompt.placeholder': 'Text to speak, or a music/SFX/voice-design description…',
   'prompt.required': 'Prompt or text is required',
   'model.label': 'Model / voice',
   'voice.label': 'Voice',