dsh-audiogen 0.3.2 → 0.3.4

package/lib/index.js

@@ -116,6 +116,7 @@ function findBase64Audio(value) {
 	const record = value;
 	for (const key of [
 		"audio",
+		"music",
 		"b64_json",
 		"base64",
 		"data",
@@ -423,15 +424,36 @@ async function minimax(channel, request, signal) {
 		}];
 	}
 	if (request.mode === "music") {
+		const MUSIC_FORMATS = /* @__PURE__ */ new Set([
+			"mp3",
+			"wav",
+			"pcm"
+		]);
+		const MUSIC_SAMPLE_RATES = /* @__PURE__ */ new Set([
+			16e3,
+			24e3,
+			32e3,
+			44100
+		]);
+		const MUSIC_BITRATES = /* @__PURE__ */ new Set([
+			32e3,
+			64e3,
+			128e3,
+			256e3
+		]);
+		const lyrics = request.lyrics?.trim() ?? "";
+		if (lyrics === "" && request.isInstrumental !== true) throw new AudioGenError("MiniMax 音乐生成需要歌词（lyrics 参数），或在「纯音乐」模式（is_instrumental=true）下生成；也可让面板/Agent 先为提示词创作一段歌词。", "lyrics-required");
 		const endpoint = `${base}/music_generation`;
 		const body = {
 			model,
 			prompt: request.prompt,
+			...lyrics === "" ? {} : { lyrics },
+			...request.isInstrumental !== void 0 ? { is_instrumental: request.isInstrumental } : {},
 			...request.duration !== void 0 ? { duration: request.duration } : {},
 			audio_setting: {
-				format: request.format ?? "mp3",
-				sample_rate: 44100,
-				bitrate: 256e3
+				format: MUSIC_FORMATS.has(request.format ?? "mp3") ? request.format ?? "mp3" : "mp3",
+				sample_rate: MUSIC_SAMPLE_RATES.has(request.sampleRate ?? 44100) ? request.sampleRate ?? 44100 : 44100,
+				bitrate: MUSIC_BITRATES.has(request.bitrate ?? 256e3) ? request.bitrate ?? 256e3 : 256e3
 			}
 		};
 		const response = await fetchWithTimeout(endpoint, {
@@ -1054,6 +1076,8 @@ function parseGenerateRequest(body) {
 		...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.lyrics === "string" && body.lyrics.trim() !== "" ? { lyrics: body.lyrics.trim() } : {},
+		...typeof body.isInstrumental === "boolean" ? { isInstrumental: body.isInstrumental } : {},
 		...typeof body.format === "string" && body.format.trim() !== "" ? { format: body.format.trim() } : {},
 		...typeof body.channelId === "string" && body.channelId !== "" ? { channelId: body.channelId } : {},
 		...str(body.emotion) !== void 0 ? { emotion: str(body.emotion) } : {},
@@ -1592,9 +1616,17 @@ function registerAgentAudioTools(ctx, resolve) {
 				type: "number",
 				description: "Requested duration in seconds for music/sfx."
 			},
+			lyrics: {
+				type: "string",
+				description: "Lyrics for music generation (MiniMax music-3.0/music-cover). Required unless is_instrumental is true. Split verses with an empty line."
+			},
+			is_instrumental: {
+				type: "boolean",
+				description: "Generate purely instrumental music without vocals/lyrics (MiniMax is_instrumental). When true, lyrics may be omitted."
+			},
 			format: {
 				type: "string",
-				description: "Output format such as mp3 or wav."
+				description: "Output format such as mp3 or wav. MiniMax music supports mp3/wav/pcm."
 			},
 			emotion: {
 				type: "string",
@@ -1623,11 +1655,11 @@ function registerAgentAudioTools(ctx, resolve) {
 			},
 			sample_rate: {
 				type: "integer",
-				description: "MiniMax TTS sample rate: 16000/24000/32000/44100/48000, default 32000 (audio_setting.sample_rate)."
+				description: "MiniMax sample rate: music 16000/24000/32000/44100 (default 44100); tts default 32000 (audio_setting.sample_rate)."
 			},
 			bitrate: {
 				type: "integer",
-				description: "MiniMax TTS bitrate in bps: 64000-320000, default 128000 (audio_setting.bitrate)."
+				description: "MiniMax bitrate in bps: 32000/64000/128000/256000 (music default 256000, tts default 128000; audio_setting.bitrate)."
 			},
 			channel: {
 				type: "integer",
@@ -1730,6 +1762,8 @@ function registerAgentAudioTools(ctx, resolve) {
 				...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.lyrics === "string" && args.lyrics.trim() !== "" ? { lyrics: args.lyrics.trim() } : {},
+				...typeof args.is_instrumental === "boolean" ? { isInstrumental: args.is_instrumental } : {},
 				...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 } : {},

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": "0.3.2",
+  "version": "0.3.4",
   "type": "module",
   "main": "lib/index.js",
   "exports": {

package/skills/music/SKILL.md

@@ -2,15 +2,38 @@
 
 ## 触发
 - `/audio:music <描述>`
-- 用户说“生成一段音乐 / 配乐 / BGM”
+- 用户说“生成一段音乐 / 配乐 / BGM / 纯音乐”
 
 ## 参数
 - prompt: 必填，风格/情绪/乐器/时长描述
-- model: 可选，已配置的音频模型
-- duration: 可选，秒数
-- format: 可选，mp3 / wav
+- model: 可选，已配置的音频模型（MiniMax：music-3.0 / music-2.6 / music-cover）
+- lyrics: 可选，歌词；**MiniMax music-3.0 / music-cover 必填**（除非 is_instrumental=true）；多段用空行分隔
+- is_instrumental: 可选，是否纯音乐（无歌词/人声），true 时 lyrics 可留空
+- duration: 可选，秒数（MiniMax 一般 5-120s）
+- format: 可选，MiniMax 音乐仅 mp3 / wav / pcm
 
 ## 流程
-1. 确认已配置支持音乐生成的渠道（如 Stability Audio / 自定义）。
-2. 调用 `generate_audio`，mode=music。
-3. 将生成的音频 URL 返回给用户。
+1. 确认已配置支持音乐生成的渠道（如 MiniMax / Stability Audio / 自定义）。
+2. 若用户未给歌词且未要求纯音乐：对 MiniMax 渠道先创作/补全一段歌词再调用。
+3. 调用 `generate_audio`，mode=music。
+4. 将生成的音频 URL 返回给用户。
+
+## MiniMax 官方 music_generation 字段参考（POST /v1/music_generation）
+
+| 字段 | 工具参数 | 说明 |
+| --- | --- | --- |
+| model | model | music-3.0 / music-2.6 / music-cover |
+| prompt | prompt | 音乐风格/情绪/乐器描述（≤3000 字） |
+| lyrics | lyrics | 歌词；多段用空行分隔；纯音乐模式可留空 |
+| is_instrumental | is_instrumental | 默认 false；true = 纯音乐（无歌词/人声），此时 lyrics 可省 |
+| duration | duration | 生成时长（秒） |
+| audio_setting.format | format | **mp3 / wav / pcm**（仅此三种） |
+| audio_setting.sample_rate | sample_rate | **16000 / 24000 / 32000 / 44100**（无 48000） |
+| audio_setting.bitrate | bitrate | **32000 / 64000 / 128000 / 256000** |
+
+> 引擎对音乐 audio_setting 按上述枚举校验，超出枚举的取值自动回退默认（format=mp3、sample_rate=44100、bitrate=256000）。
+> 若不加 lyrics 也未开启纯音乐，引擎会直接提示 `lyrics-required`（MiniMax 上游返回 2013 lyrics is required）。
+
+## 常见错误
+- `lyrics-required`：MiniMax 音乐生成需要歌词，或开启纯音乐。
+- `HTTP 400` 且含 `2013`：上游参数不合法，检查 lyrics / audio_setting 枚举。

package/src/agent-audio-tools.ts

@@ -109,7 +109,9 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
       preview_text: { type: 'string', description: 'Optional preview text for voice_design.' },
       speed: { type: 'number', description: 'Optional speaking rate / speed multiplier where supported. MiniMax range 0.5-2.0 (default 1).' },
       duration: { type: 'number', description: 'Requested duration in seconds for music/sfx.' },
-      format: { type: 'string', description: 'Output format such as mp3 or wav.' },
+      lyrics: { type: 'string', description: 'Lyrics for music generation (MiniMax music-3.0/music-cover). Required unless is_instrumental is true. Split verses with an empty line.' },
+      is_instrumental: { type: 'boolean', description: 'Generate purely instrumental music without vocals/lyrics (MiniMax is_instrumental). When true, lyrics may be omitted.' },
+      format: { type: 'string', description: 'Output format such as mp3 or wav. MiniMax music supports mp3/wav/pcm.' },
       // ---- MiniMax TTS only (ignored by other providers) ----
       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).' },
@@ -117,8 +119,8 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
       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).' },
+      sample_rate: { type: 'integer', description: 'MiniMax sample rate: music 16000/24000/32000/44100 (default 44100); tts default 32000 (audio_setting.sample_rate).' },
+      bitrate: { type: 'integer', description: 'MiniMax bitrate in bps: 32000/64000/128000/256000 (music default 256000, tts 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).' },
@@ -195,6 +197,8 @@ export function registerAgentAudioTools(ctx: Context, resolve: () => AgentAudioT
         ...(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.lyrics === 'string' && args.lyrics.trim() !== '' ? { lyrics: args.lyrics.trim() } : {}),
+        ...(typeof args.is_instrumental === 'boolean' ? { isInstrumental: args.is_instrumental } : {}),
         ...(typeof args.format === 'string' && args.format.trim() !== '' ? { format: args.format.trim() } : {}),
         // ---- MiniMax TTS 专属字段 ----
         ...(typeof args.emotion === 'string' && args.emotion.trim() !== '' ? { emotion: args.emotion.trim() } : {}),

package/src/audio-engine.ts

@@ -124,7 +124,7 @@ function findBase64Audio(value: unknown): string | undefined {
   }
   if (value === null || typeof value !== 'object') return undefined
   const record = value as Record<string, unknown>
-  for (const key of ['audio', 'b64_json', 'base64', 'data', 'output', 'result', 'value']) {
+  for (const key of ['audio', 'music', 'b64_json', 'base64', 'data', 'output', 'result', 'value']) {
     const candidate = record[key]
     const found = findBase64Audio(candidate)
     if (found !== undefined) return found
@@ -437,15 +437,31 @@ async function minimax(channel: AudioChannel, request: GenerateAudioRequest, sig
   }
 
   if (request.mode === 'music') {
+    // MiniMax 音乐生成官方字段：model/lyrics/prompt/is_instrumental/duration/
+    // audio_setting{format, sample_rate, bitrate}。音频输出配置为固定枚举：
+    // format mp3|wav|pcm；sample_rate 16000|24000|32000|44100；
+    // bitrate 32000|64000|128000|256000，超出枚举的值回退默认。
+    const MUSIC_FORMATS = new Set(['mp3', 'wav', 'pcm'])
+    const MUSIC_SAMPLE_RATES = new Set([16000, 24000, 32000, 44100])
+    const MUSIC_BITRATES = new Set([32000, 64000, 128000, 256000])
+    const lyrics = request.lyrics?.trim() ?? ''
+    if (lyrics === '' && request.isInstrumental !== true) {
+      throw new AudioGenError(
+        'MiniMax 音乐生成需要歌词（lyrics 参数），或在「纯音乐」模式（is_instrumental=true）下生成；也可让面板/Agent 先为提示词创作一段歌词。',
+        'lyrics-required',
+      )
+    }
     const endpoint = `${base}/music_generation`
     const body: Record<string, unknown> = {
       model,
       prompt: request.prompt,
+      ...(lyrics === '' ? {} : { lyrics }),
+      ...(request.isInstrumental !== undefined ? { is_instrumental: request.isInstrumental } : {}),
       ...(request.duration !== undefined ? { duration: request.duration } : {}),
       audio_setting: {
-        format: request.format ?? 'mp3',
-        sample_rate: 44100,
-        bitrate: 256000,
+        format: MUSIC_FORMATS.has(request.format ?? 'mp3') ? (request.format ?? 'mp3') : 'mp3',
+        sample_rate: MUSIC_SAMPLE_RATES.has(request.sampleRate ?? 44100) ? (request.sampleRate ?? 44100) : 44100,
+        bitrate: MUSIC_BITRATES.has(request.bitrate ?? 256000) ? (request.bitrate ?? 256000) : 256000,
       },
     }
     const response = await fetchWithTimeout(endpoint, {

package/src/client/AudioGenPanel.tsx

@@ -60,6 +60,8 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
   const [voice, setVoice] = useState('')
   const [speed, setSpeed] = useState('')
   const [duration, setDuration] = useState('')
+  const [lyrics, setLyrics] = useState('')
+  const [instrumental, setInstrumental] = useState(false)
   const [format, setFormat] = useState('mp3')
   // MiniMax TTS 高级参数（其他厂商忽略）
   const [emotion, setEmotion] = useState('')
@@ -109,6 +111,8 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
         ...(voice.trim() !== '' ? { voice: voice.trim() } : {}),
         ...(speed.trim() !== '' ? { speed: Number(speed) } : {}),
         ...(duration.trim() !== '' ? { duration: Number(duration) } : {}),
+        ...(lyrics.trim() !== '' ? { lyrics: lyrics.trim() } : {}),
+        ...(instrumental ? { isInstrumental: true } : {}),
         ...(format.trim() !== '' ? { format: format.trim() } : {}),
         ...(emotion.trim() !== '' ? { emotion: emotion.trim() } : {}),
         ...(vol.trim() !== '' ? { vol: Number(vol) } : {}),
@@ -252,14 +256,53 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
             </label>
           ) : null}
 
+          {mode === 'music' ? (
+            <>
+              <label className={css.label}>
+                <span>歌词（纯音乐模式可留空；多段用空行分隔）</span>
+                <textarea className={css.textarea} value={lyrics} onChange={event => setLyrics(event.target.value)} placeholder={'第一段歌词…\n\n第二段歌词…'} />
+              </label>
+              <label className={css.checkbox}>
+                <input type="checkbox" checked={instrumental} onChange={event => setInstrumental(event.target.checked)} />
+                <span>纯音乐（无歌词/人声）is_instrumental</span>
+              </label>
+              <div className={css.row}>
+                <label className={css.label}>
+                  <span>采样率</span>
+                  <select className={css.select} value={sampleRate} onChange={event => setSampleRate(event.target.value)}>
+                    <option value="">默认（44100）</option>
+                    <option value="16000">16000</option>
+                    <option value="24000">24000</option>
+                    <option value="32000">32000</option>
+                    <option value="44100">44100</option>
+                  </select>
+                </label>
+                <label className={css.label}>
+                  <span>码率 bps</span>
+                  <select className={css.select} value={bitrate} onChange={event => setBitrate(event.target.value)}>
+                    <option value="">默认（256000）</option>
+                    <option value="32000">32000</option>
+                    <option value="64000">64000</option>
+                    <option value="128000">128000</option>
+                    <option value="256000">256000</option>
+                  </select>
+                </label>
+              </div>
+            </>
+          ) : 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>
+                {mode === 'tts' ? (
+                  <>
+                    <option value="flac">flac</option>
+                    <option value="ogg">ogg</option>
+                  </>
+                ) : null}
                 <option value="pcm">pcm</option>
               </select>
             </label>

package/src/client/locales.ts

@@ -12,7 +12,7 @@ export const zh = {
   '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',
+  'model.label': '模型',
   'voice.label': '音色',
   'speed.label': '语速',
   'duration.label': '时长（秒）',
@@ -113,7 +113,7 @@ export const en: Record<AudioGenKey, string> = {
   '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',
+  'model.label': 'Model',
   'voice.label': 'Voice',
   'speed.label': 'Speed',
   'duration.label': 'Duration (s)',

package/src/protocol.ts

@@ -8,7 +8,7 @@
 export const AUDIOGEN_SETTINGS_NAMESPACE = 'dsh-audiogen'
 
 /** Published package version shared by the host updater and the client UI. */
-export const PLUGIN_VERSION = '0.3.2'
+export const PLUGIN_VERSION = '0.3.4'
 
 /** Same-origin route family (loopback-only, mirroring dsh-imagegen). */
 export const SETTINGS_API = {
@@ -114,6 +114,10 @@ export interface GenerateAudioRequest {
   speed?: number
   /** Requested duration in seconds (music/sfx). */
   duration?: number
+  /** MiniMax 音乐生成歌词；music-3.0 / music-cover 在非纯音乐模式下必填。 */
+  lyrics?: string
+  /** MiniMax 是否生成纯音乐（无歌词/人声）；true 时 lyrics 可为空。 */
+  isInstrumental?: boolean
   /** Output format, e.g. mp3, wav, pcm. */
   format?: string
   /** Channel this request targets (host falls back to default). */

package/src/routes.ts

@@ -122,6 +122,8 @@ function parseGenerateRequest(body: Record<string, unknown>): GenerateAudioReque
     ...(typeof body.previewText === 'string' && body.previewText.trim() !== '' ? { previewText: body.previewText.trim() } : {}),
     ...(num(body.speed) !== undefined ? { speed: num(body.speed)! } : {}),
     ...(num(body.duration) !== undefined ? { duration: num(body.duration)! } : {}),
+    ...(typeof body.lyrics === 'string' && body.lyrics.trim() !== '' ? { lyrics: body.lyrics.trim() } : {}),
+    ...(typeof body.isInstrumental === 'boolean' ? { isInstrumental: body.isInstrumental } : {}),
     ...(typeof body.format === 'string' && body.format.trim() !== '' ? { format: body.format.trim() } : {}),
     ...(typeof body.channelId === 'string' && body.channelId !== '' ? { channelId: body.channelId } : {}),
     // ---- MiniMax TTS 专属字段（其他厂商忽略） ----