dsh-audiogen 0.3.0 → 0.3.2

package/src/audio-presets.ts

@@ -1,6 +1,10 @@
 /**
  * Built-in audio provider catalog (presets).
  * Framework-free pure data; served to the settings card through a host route.
+ *
+ * Only the officially supported audio vendors are offered here — MiniMax,
+ * ElevenLabs and Stability AI. Any other endpoint (including OpenAI-compatible
+ * TTS gateways) is added through the "自定义渠道" flow instead.
  */
 
 import type { ModelMapping } from './protocol.ts'
@@ -15,39 +19,19 @@ export interface AudioPresetProvider {
   apiUrl: string
   /** One-line description shown in the picker. */
   hint: string
+  /** Official vendor website, shown as a link in the channel editor. */
+  site?: string
   /** Known model/voice catalog prefilled into the channel. */
   models: ModelMapping[]
 }
 
 export const AUDIO_PRESETS: AudioPresetProvider[] = [
-  {
-    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: [
       // TTS models
       { alias: 'speech-2.8-hd', id: 'speech-2.8-hd', category: 'tts' },
@@ -64,23 +48,33 @@ export const AUDIO_PRESETS: AudioPresetProvider[] = [
       { alias: 'music-cover', id: 'music-cover', category: 'music' },
     ],
   },
+  {
+    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', category: 'music' },
       { alias: 'stable-audio-1.0', 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. */

package/src/client/AudioGenPanel.tsx

@@ -61,11 +61,25 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
   const [speed, setSpeed] = useState('')
   const [duration, setDuration] = useState('')
   const [format, setFormat] = useState('mp3')
+  // MiniMax TTS 高级参数（其他厂商忽略）
+  const [emotion, setEmotion] = useState('')
+  const [vol, setVol] = useState('')
+  const [pitch, setPitch] = useState('')
+  const [toneText, setToneText] = useState('')
+  const [sampleRate, setSampleRate] = useState('')
+  const [bitrate, setBitrate] = useState('')
+  const [audioChannel, setAudioChannel] = useState('')
+  const [subtitle, setSubtitle] = useState(false)
   const [loading, setLoading] = useState(false)
   const [error, setError] = useState<string | null>(null)
   const [outputs, setOutputs] = useState<GeneratedAudio[]>([])
   const { entries, reload, clear } = useHistory()
 
+  const isMiniMaxChannel = useMemo(() => {
+    const target = channels.find(candidate => candidate.id === modelOptions.defaultChannelId) ?? channels[0]
+    return target !== undefined && (target.preset === 'minimax' || /minimax/i.test(target.apiUrl))
+  }, [channels, modelOptions.defaultChannelId])
+
   const visibleModels = useMemo(() => {
     if (mode === 'voice_design') return []
     return modelOptions.models
@@ -96,6 +110,14 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
         ...(speed.trim() !== '' ? { speed: Number(speed) } : {}),
         ...(duration.trim() !== '' ? { duration: Number(duration) } : {}),
         ...(format.trim() !== '' ? { format: format.trim() } : {}),
+        ...(emotion.trim() !== '' ? { emotion: emotion.trim() } : {}),
+        ...(vol.trim() !== '' ? { vol: Number(vol) } : {}),
+        ...(pitch.trim() !== '' ? { pitch: Number(pitch) } : {}),
+        ...(toneText.trim() !== '' ? { pronunciationTone: toneText.split('\n').map(item => item.trim()).filter(item => item !== '') } : {}),
+        ...(sampleRate.trim() !== '' ? { sampleRate: Number(sampleRate) } : {}),
+        ...(bitrate.trim() !== '' ? { bitrate: Number(bitrate) } : {}),
+        ...(audioChannel.trim() !== '' ? { audioChannel: Number(audioChannel) } : {}),
+        ...(subtitle ? { subtitleEnable: true } : {}),
       })
       if (!response.ok) {
         setError(response.message ?? '生成失败')
@@ -166,7 +188,7 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
           {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" />
+              <input className={css.input} value={voice} onChange={event => setVoice(event.target.value)} placeholder={isMiniMaxChannel ? 'male-qn-qingse / female-shaonv' : 'alloy / 自定义音色'} />
             </label>
           ) : null}
 
@@ -177,6 +199,52 @@ export function AudioGenPanel(props: { api: AudiogenApi; scope: AudiogenScope })
             </label>
           ) : null}
 
+          {mode === 'tts' && isMiniMaxChannel ? (
+            <details className={css.advanced}>
+              <summary>MiniMax 高级参数</summary>
+              <label className={css.label}>
+                <span>情绪 emotion</span>
+                <input className={css.input} value={emotion} onChange={event => setEmotion(event.target.value)} placeholder="happy / sad / angry / nervous…" />
+              </label>
+              <div className={css.row}>
+                <label className={css.label}>
+                  <span>音量 vol (0-10)</span>
+                  <input className={css.input} type="number" min="0" max="10" step="0.5" value={vol} onChange={event => setVol(event.target.value)} placeholder="1" />
+                </label>
+                <label className={css.label}>
+                  <span>音调 pitch (-12~12)</span>
+                  <input className={css.input} type="number" min="-12" max="12" value={pitch} onChange={event => setPitch(event.target.value)} placeholder="0" />
+                </label>
+              </div>
+              <div className={css.row}>
+                <label className={css.label}>
+                  <span>采样率</span>
+                  <input className={css.input} type="number" min="16000" max="48000" step="8000" value={sampleRate} onChange={event => setSampleRate(event.target.value)} placeholder="32000" />
+                </label>
+                <label className={css.label}>
+                  <span>码率 bps</span>
+                  <input className={css.input} type="number" min="64000" max="320000" step="8000" value={bitrate} onChange={event => setBitrate(event.target.value)} placeholder="128000" />
+                </label>
+                <label className={css.label}>
+                  <span>声道</span>
+                  <select className={css.select} value={audioChannel} onChange={event => setAudioChannel(event.target.value)}>
+                    <option value="">默认(1)</option>
+                    <option value="1">1</option>
+                    <option value="2">2</option>
+                  </select>
+                </label>
+              </div>
+              <label className={css.label}>
+                <span>发音词典（每行一条："文字/读音"）</span>
+                <textarea className={css.textarea} value={toneText} onChange={event => setToneText(event.target.value)} placeholder={'处理/(chu3)(li3)\n危险/dangerous'} />
+              </label>
+              <label className={css.checkbox}>
+                <input type="checkbox" checked={subtitle} onChange={event => setSubtitle(event.target.checked)} />
+                <span>生成字幕 subtitle_enable</span>
+              </label>
+            </details>
+          ) : null}
+
           {mode === 'music' || mode === 'sfx' ? (
             <label className={css.label}>
               <span>{tt('duration.label')}</span>