dsh-audiogen 0.4.1 → 0.4.2

package/src/client/audio-panel.module.css

@@ -425,6 +425,116 @@
   background: color-mix(in srgb, var(--dsw-alias-brand-primary, #2563eb) 8%, transparent);
 }
 
+/* ---------------------------------------------------------- model compare */
+
+.compareBox {
+  display: flex;
+  flex-direction: column;
+  gap: 7px;
+  padding: 9px 11px;
+  border: 1px dashed var(--dsw-alias-brand-primary, #2563eb);
+  border-radius: 10px;
+  background: color-mix(in srgb, var(--dsw-alias-brand-primary, #2563eb) 4%, var(--dsw-alias-bg-layer-1, #fff));
+}
+
+.compareChips {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+}
+
+.compareChip {
+  display: inline-flex;
+  align-items: center;
+  min-height: 27px;
+  padding: 0 11px;
+  border: 1px solid var(--dsw-alias-border-l2, #d1d5db);
+  border-radius: 999px;
+  background: var(--dsw-alias-bg-layer-1, #fff);
+  color: var(--dsw-alias-label-secondary, #6b7280);
+  font-family: inherit;
+  font-size: 12px;
+  cursor: pointer;
+  transition: color 0.12s, border-color 0.12s, background 0.12s;
+}
+
+.compareChip:hover {
+  color: var(--dsw-alias-label-primary, #1f2328);
+  border-color: var(--dsw-alias-label-dimmed, #9ca3af);
+}
+
+.compareChip[data-active='true'] {
+  color: var(--dsw-alias-brand-primary, #2563eb);
+  border-color: var(--dsw-alias-brand-primary, #2563eb);
+  background: color-mix(in srgb, var(--dsw-alias-brand-primary, #2563eb) 9%, transparent);
+  font-weight: 600;
+}
+
+.compareBoard {
+  display: flex;
+  flex-direction: column;
+  gap: 12px;
+}
+
+.compareGroup {
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+  padding: 11px;
+  border: 1px solid var(--dsw-alias-border-l1, #e5e7eb);
+  border-radius: 12px;
+  background: var(--dsw-alias-bg-layer-2, #f7f7f8);
+}
+
+.compareGroup[data-state='running'] {
+  border-color: color-mix(in srgb, var(--dsw-alias-brand-primary, #2563eb) 45%, var(--dsw-alias-border-l1, #e5e7eb));
+}
+
+.compareGroup[data-state='done'] {
+  border-color: color-mix(in srgb, #10b981 45%, var(--dsw-alias-border-l1, #e5e7eb));
+}
+
+.compareGroup[data-state='error'] {
+  border-color: color-mix(in srgb, var(--dsw-alias-label-error, #b91c1c) 45%, var(--dsw-alias-border-l1, #e5e7eb));
+}
+
+.compareGroupHead {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: 10px;
+}
+
+.compareModelName {
+  overflow: hidden;
+  padding: 3px 10px;
+  border-radius: 999px;
+  background: var(--dsw-alias-bg-layer-1, #fff);
+  border: 1px solid var(--dsw-alias-border-l1, #e5e7eb);
+  color: var(--dsw-alias-label-primary, #1f2328);
+  font-size: 12px;
+  font-weight: 600;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.compareState {
+  display: inline-flex;
+  align-items: center;
+  gap: 5px;
+  font-size: 11px;
+  color: var(--dsw-alias-label-tertiary, #9ca3af);
+  white-space: nowrap;
+}
+
+.compareGroup[data-state='running'] .compareState { color: var(--dsw-alias-brand-primary, #2563eb); }
+.compareGroup[data-state='done'] .compareState { color: #059669; }
+.compareGroup[data-state='error'] .compareState { color: var(--dsw-alias-label-error, #b91c1c); }
+
+.hint[data-error] {
+  color: var(--dsw-alias-label-error, #b91c1c);
+}
+
 /* --------------------------------------------------------------- history */
 
 .historyCol {
@@ -892,3 +1002,134 @@
   color: var(--dsw-alias-label-primary, #1f2328);
   white-space: normal;
 }
+
+/* 任务列表 */
+.taskList {
+  display: flex;
+  flex-direction: column;
+  gap: 14px;
+}
+
+.taskCard {
+  border: 1px solid var(--dsw-alias-border-l2, #d1d5db);
+  border-radius: 10px;
+  padding: 10px 12px;
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+  background: var(--dsw-alias-bg-layer-2, #fafafa);
+}
+
+.taskCard[data-state='failed'] {
+  border-color: #dc2626;
+}
+
+.taskCard[data-state='cancelled'] {
+  opacity: 0.75;
+}
+
+.taskHead {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  flex-wrap: wrap;
+}
+
+.taskLabel {
+  font-size: 13px;
+  font-weight: 600;
+  color: var(--dsw-alias-label-primary, #1f2328);
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+  max-width: 200px;
+}
+
+.taskStatus {
+  font-size: 12px;
+  color: var(--dsw-alias-label-secondary, #6b7280);
+}
+
+.taskStatus[data-state='done'] {
+  color: #16a34a;
+}
+
+.taskStatus[data-state='failed'] {
+  color: #dc2626;
+}
+
+.taskActions {
+  margin-left: auto;
+  display: flex;
+  gap: 6px;
+}
+
+/* 历史记录分类 */
+.historyTabs {
+  display: flex;
+  flex-wrap: wrap;
+  gap: 6px;
+  margin-bottom: 10px;
+}
+
+.historyTab {
+  border: 1px solid var(--dsw-alias-border-l2, #d1d5db);
+  border-radius: 999px;
+  padding: 3px 10px;
+  font-size: 12px;
+  cursor: pointer;
+  background: var(--dsw-alias-bg-layer-3, #fff);
+  color: var(--dsw-alias-label-secondary, #6b7280);
+}
+
+.historyTab[data-active='true'] {
+  background: var(--dsw-alias-bg-hover, #f3f4f6);
+  color: var(--dsw-alias-label-primary, #1f2328);
+  font-weight: 600;
+}
+
+.historyTabCount {
+  margin-left: 4px;
+  font-size: 11px;
+  color: var(--dsw-alias-label-tertiary, #9ca3af);
+}
+
+/* 对比任务卡 */
+.historyCompareSummary {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  cursor: pointer;
+  flex-wrap: wrap;
+}
+
+.historyCompareBadge {
+  font-size: 11px;
+  font-weight: 600;
+  color: #7c3aed;
+  border: 1px solid #ddd6fe;
+  border-radius: 999px;
+  padding: 1px 8px;
+  background: #f5f3ff;
+  white-space: nowrap;
+}
+
+.historyModelRow {
+  margin-top: 8px;
+  padding-top: 8px;
+  border-top: 1px dashed var(--dsw-alias-border-l1, #e5e7eb);
+  display: flex;
+  flex-direction: column;
+  gap: 4px;
+}
+
+.overrideOnly {
+  font-size: 10px;
+  color: var(--dsw-alias-label-tertiary, #9ca3af);
+  font-weight: 500;
+}
+
+.overrideDash {
+  color: var(--dsw-alias-label-tertiary, #9ca3af);
+  padding: 0 6px;
+}

package/src/client/field-specs.ts

@@ -0,0 +1,186 @@
+/**
+ * 厂商感知的字段规格表：按（渠道 preset × 模式）声明面板该显示哪些参数，
+ * 以及每个参数适用于哪些渠道。引擎不变——引擎本来就按渠道消费字段；
+ * 这里只负责「该显示什么、标签/枚举/校验按渠道」。
+ */
+
+import type { AudioMode } from '../protocol.ts'
+
+/** 与面板表单 state 对应的一组字段键。 */
+export type FieldKey =
+  | 'duration'
+  | 'format'
+  | 'lyrics'
+  | 'instrumental'
+  | 'sampleRate'
+  | 'bitrate'
+  | 'audioChannel'
+  | 'voice'
+  | 'speed'
+  | 'emotion'
+  | 'vol'
+  | 'pitch'
+  | 'toneText'
+  | 'subtitle'
+  | 'loop'
+  | 'promptInfluence'
+  | 'seed'
+  | 'steps'
+  | 'cfgScale'
+
+export interface FieldSpec {
+  key: FieldKey
+  label: string
+  type: 'text' | 'number' | 'select' | 'checkbox'
+  options?: string[]
+  placeholder?: string
+  min?: number
+  max?: number
+  step?: number
+  /** 官方字段名/说明（tooltip）。 */
+  hint?: string
+  /** 放在折叠「高级参数」里。 */
+  advanced?: boolean
+  /** 适用的渠道 preset；undefined 表示通用。 */
+  presets?: string[]
+}
+
+const PRESET_MINIMAX = 'minimax'
+const PRESET_ELEVENLABS = 'elevenlabs'
+const PRESET_STABILITY = 'stability-audio'
+
+/** 各 preset 支持的音乐输出格式（交集用于全局字段）。 */
+const MUSIC_FORMATS: Record<string, string[]> = {
+  [PRESET_MINIMAX]: ['mp3', 'wav', 'pcm'],
+  [PRESET_ELEVENLABS]: ['mp3', 'wav'],
+  [PRESET_STABILITY]: ['mp3', 'wav'],
+}
+
+const MUSIC_KEYS: FieldKey[] = ['duration', 'format', 'lyrics', 'instrumental', 'sampleRate', 'bitrate']
+const TTS_KEYS: FieldKey[] = ['voice', 'speed', 'format', 'emotion', 'vol', 'pitch', 'toneText', 'sampleRate', 'bitrate', 'audioChannel', 'subtitle']
+const SFX_KEYS: FieldKey[] = ['duration', 'format', 'loop', 'promptInfluence', 'seed', 'steps', 'cfgScale']
+
+export function presetSupports(preset: string, key: FieldKey, mode: AudioMode): boolean {
+  const p = preset.toLowerCase()
+  const K = MUSIC_KEYS.includes(key) ? MUSIC_KEYS : TTS_KEYS.includes(key) ? TTS_KEYS : SFX_KEYS
+  if (!K.includes(key)) return true
+  switch (key) {
+    case 'lyrics':
+    case 'instrumental':
+      return p === PRESET_MINIMAX || p === PRESET_ELEVENLABS
+    case 'sampleRate':
+    case 'bitrate':
+    case 'audioChannel':
+      return p === PRESET_MINIMAX
+    case 'emotion':
+    case 'vol':
+    case 'pitch':
+    case 'toneText':
+    case 'subtitle':
+      return p === PRESET_MINIMAX
+    case 'loop':
+    case 'promptInfluence':
+      return p === PRESET_ELEVENLABS
+    case 'seed':
+    case 'steps':
+    case 'cfgScale':
+      return p === PRESET_STABILITY
+    default:
+      return true
+  }
+}
+
+const SPECS: Record<FieldKey, Omit<FieldSpec, 'key' | 'presets'>> = {
+  duration: { label: '时长（秒）', type: 'number', min: 1, max: 120, placeholder: '30', hint: 'duration；MiniMax 音乐 ≤190、ElevenLabs 音乐 3-600（转 ms）、Stability 按模型 190/380' },
+  format: { label: '输出格式', type: 'select', options: ['mp3', 'wav', 'pcm', 'flac', 'ogg'], hint: 'format / output_format / response_format' },
+  lyrics: { label: '歌词（纯音乐模式可留空；多段用空行分隔）', type: 'text', placeholder: '第一段歌词…\n\n第二段歌词…', hint: 'MiniMax lyrics / ElevenLabs lyrics_text' },
+  instrumental: { label: '纯音乐（无歌词/人声）', type: 'checkbox', hint: 'MiniMax is_instrumental / ElevenLabs force_instrumental' },
+  sampleRate: { label: '采样率', type: 'select', options: ['16000', '24000', '32000', '44100'], placeholder: '默认（44100）', hint: 'audio_setting.sample_rate：16000-44100' },
+  bitrate: { label: '码率 bps', type: 'select', options: ['32000', '64000', '128000', '256000'], placeholder: '默认（256000）', hint: 'audio_setting.bitrate：32000-256000' },
+  audioChannel: { label: '声道', type: 'select', options: ['1', '2'], placeholder: '默认(1)', hint: 'audio_setting.channel（TTS）' },
+  voice: { label: '音色', type: 'text', placeholder: '自定义音色', hint: 'voice_id（MiniMax 必填）/ voice（ElevenLabs）' },
+  speed: { label: '语速', type: 'number', min: 0.5, max: 2, step: 0.1, placeholder: '1.0', hint: 'speed（MiniMax 0.5-2）' },
+  emotion: { label: '情绪 emotion', type: 'text', placeholder: 'happy / sad / angry / nervous…', hint: 'voice_setting.emotion', advanced: true },
+  vol: { label: '音量 vol (0-10)', type: 'number', min: 0, max: 10, step: 0.5, placeholder: '1', hint: 'voice_setting.vol', advanced: true },
+  pitch: { label: '音调 pitch (-12~12)', type: 'number', min: -12, max: 12, placeholder: '0', hint: 'voice_setting.pitch', advanced: true },
+  toneText: { label: '发音词典（每行一条："文字/读音"）', type: 'text', placeholder: '处理/(chu3)(li3)\n危险/dangerous', hint: 'pronunciation_dict.tone', advanced: true },
+  subtitle: { label: '生成字幕 subtitle_enable', type: 'checkbox', hint: 'subtitle_enable', advanced: true },
+  loop: { label: '循环音效 loop（无缝循环）', type: 'checkbox', hint: 'ElevenLabs loop（仅 eleven_text_to_sound_v2）' },
+  promptInfluence: { label: '提示词影响度 prompt_influence (0-1)', type: 'number', min: 0, max: 1, step: 0.1, placeholder: '0.3', hint: 'prompt_influence：越高越贴提示词' },
+  seed: { label: 'seed', type: 'number', min: 0, max: 4294967294, placeholder: '默认（随机）', hint: 'Stable Audio seed（同参数可复现）' },
+  steps: { label: 'steps', type: 'number', min: 4, max: 100, placeholder: '默认', hint: 'Stable Audio 采样步数（2: 30-100；2.5/3: 4-8）' },
+  cfgScale: { label: 'cfg_scale', type: 'number', min: 1, max: 25, placeholder: '默认', hint: 'Stable Audio 提示词遵循度（2 默认 7，2.5/3 默认 1）' },
+}
+
+function specOf(key: FieldKey, presets?: string[]): FieldSpec {
+  return { key, ...SPECS[key], ...(presets === undefined ? {} : { presets }) }
+}
+
+/** 当前模式 + 所选模型集合（渠道集合）对应的「全局字段」清单。 */
+export function globalFieldSpecs(mode: AudioMode, presets: string[]): FieldSpec[] {
+  const list: FieldSpec[] = []
+  const all = (key: FieldKey, keys: FieldKey[]): boolean => keys.includes(key) && presets.every(preset => presetSupports(preset, key, mode))
+  if (mode === 'tts') {
+    list.push(specOf('voice'), specOf('speed'))
+    list.push({ ...specOf('format'), options: ['mp3', 'wav', 'flac', 'ogg', 'pcm'] })
+    const advanced = (['emotion', 'vol', 'pitch', 'toneText', 'sampleRate', 'bitrate', 'audioChannel', 'subtitle'] as FieldKey[]).filter(key => all(key, TTS_KEYS) && presets.length > 0)
+    for (const key of advanced) list.push(specOf(key))
+  } else if (mode === 'music') {
+    list.push(specOf('duration'))
+    // 格式选项取所有选中渠道支持的交集（无交集则回退 mp3/wav）
+    const supported: string[][] = presets.length === 0 ? [['mp3', 'wav', 'pcm']] : presets.map(preset => MUSIC_FORMATS[preset.toLowerCase()] ?? ['mp3', 'wav'])
+    const intersect: string[] = supported.reduce<string[]>((acc, cur) => acc.filter(item => cur.includes(item)), supported[0] ?? ['mp3', 'wav'])
+    list.push({ ...specOf('format'), options: intersect.length > 0 ? intersect : ['mp3', 'wav'] })
+    if (all('lyrics', MUSIC_KEYS) && presets.length > 0) list.push(specOf('lyrics'))
+    if (all('instrumental', MUSIC_KEYS) && presets.length > 0) list.push(specOf('instrumental'))
+    if (all('sampleRate', MUSIC_KEYS) && presets.length > 0) list.push(specOf('sampleRate'))
+    if (all('bitrate', MUSIC_KEYS) && presets.length > 0) list.push(specOf('bitrate'))
+  } else if (mode === 'sfx') {
+    list.push(specOf('duration'))
+    list.push({ ...specOf('format'), options: ['mp3', 'wav', 'pcm'] })
+    if (all('loop', SFX_KEYS) && presets.length > 0) list.push(specOf('loop'))
+    if (all('promptInfluence', SFX_KEYS) && presets.length > 0) list.push(specOf('promptInfluence'))
+    if (all('seed', SFX_KEYS) && presets.length > 0) list.push(specOf('seed'), specOf('steps'), specOf('cfgScale'))
+  }
+  return list
+}
+
+/** 「每模型参数覆盖」矩阵的字段全集（含适用渠道标注）。 */
+export function overrideRowSpecs(mode: AudioMode): Array<FieldSpec & { presets: string[] }> {
+  const rows: Array<FieldSpec & { presets: string[] }> = []
+  const presets: Array<[FieldKey, string[]]> = [
+    ['format', [PRESET_MINIMAX, PRESET_ELEVENLABS, PRESET_STABILITY]],
+    ['duration', [PRESET_MINIMAX, PRESET_ELEVENLABS, PRESET_STABILITY]],
+    ['voice', [PRESET_MINIMAX, PRESET_ELEVENLABS]],
+    ['speed', [PRESET_MINIMAX, PRESET_ELEVENLABS]],
+    ['lyrics', [PRESET_MINIMAX, PRESET_ELEVENLABS]],
+    ['instrumental', [PRESET_MINIMAX, PRESET_ELEVENLABS]],
+    ['sampleRate', [PRESET_MINIMAX]],
+    ['bitrate', [PRESET_MINIMAX]],
+    ['audioChannel', [PRESET_MINIMAX]],
+    ['emotion', [PRESET_MINIMAX]],
+    ['vol', [PRESET_MINIMAX]],
+    ['pitch', [PRESET_MINIMAX]],
+    ['toneText', [PRESET_MINIMAX]],
+    ['subtitle', [PRESET_MINIMAX]],
+    ['loop', [PRESET_ELEVENLABS]],
+    ['promptInfluence', [PRESET_ELEVENLABS]],
+    ['seed', [PRESET_STABILITY]],
+    ['steps', [PRESET_STABILITY]],
+    ['cfgScale', [PRESET_STABILITY]],
+  ]
+  for (const [key, applicable] of presets) {
+    const spec = specOf(key, applicable)
+    rows.push({ ...spec, presets: applicable })
+  }
+  return rows
+}
+
+/** 渠道 preset 的展示名。 */
+export function presetLabel(preset: string): string {
+  if (preset === PRESET_MINIMAX) return 'MiniMax'
+  if (preset === PRESET_ELEVENLABS) return 'ElevenLabs'
+  if (preset === PRESET_STABILITY) return 'Stability'
+  if (preset === 'openai-tts') return 'OpenAI'
+  return '自定义'
+}

package/src/client/library-view.tsx

@@ -319,7 +319,12 @@ export function LibraryView(props: {
       </div>
 
       {loading ? <p className={css.stateNote}>加载中…</p> : null}
-      {error !== null ? <p className={css.stateNote} data-error>{error}</p> : null}
+      {error !== null ? (
+        <div className={css.stateNote} data-error>
+          <p>{error}</p>
+          <p>若提示 not found / 404：插件宿主尚未加载新代码，请重启 `dsh web` 后在浏览器强制刷新（Cmd+Shift+R）。</p>
+        </div>
+      ) : null}
       {!loading && error === null && filtered.length === 0 ? (
         <div className={css.empty}>
           <span className={css.emptyIcon}>🎧</span>

package/src/client/locales.ts

@@ -36,6 +36,7 @@ export const zh = {
   'settings.announceToAgent': '向 Agent 播报本插件',
   'settings.allowAgentAudio': '允许 Agent 调用音频生成',
   'settings.autoSaveLibrary': '生成后自动保存到资源库',
+  'settings.maxConcurrent': '最大并发生成数（同时打到上游的请求数，默认 5）',
   'settings.save': '保存',
   'settings.saving': '保存中…',
   'settings.discard': '放弃修改',
@@ -140,6 +141,7 @@ export const en: Record<AudioGenKey, string> = {
   'settings.announceToAgent': 'Announce this plugin to agents',
   'settings.allowAgentAudio': 'Allow agents to generate audio',
   'settings.autoSaveLibrary': 'Auto-save generated audio to the library',
+  'settings.maxConcurrent': 'Max concurrent generations (in-flight upstream calls, default 5)',
   'settings.save': 'Save',
   'settings.saving': 'Saving…',
   'settings.discard': 'Discard',

package/src/client/settings-scope.ts

@@ -271,8 +271,17 @@ export function bindAudiogenScope(fetchFn: typeof fetch = fetch): AudiogenScope
  * Falls back to the legacy flat allow-list while no channels exist (upgrade
  * path). Pure projection — no host calls.
  */
+export interface ModelOption {
+  alias: string
+  category?: AudioModelCategory
+  /** 所属渠道 id/名称/preset（面板按渠道渲染参数与分组下拉）。 */
+  channelId: string
+  channelName: string
+  preset: string
+}
+
 export function audioModelOptions(config: AudiogenConfig | undefined): {
-  models: Array<{ alias: string; category?: AudioModelCategory }>
+  models: ModelOption[]
   defaultChannelId?: string
 } {
   const channels = config?.channels ?? []
@@ -283,14 +292,20 @@ export function audioModelOptions(config: AudiogenConfig | undefined): {
     ? config.defaultChannelId
     : channels[0]!.id
   const ordered = [defaultId, ...channels.filter(channel => channel.id !== defaultId).map(channel => channel.id)]
-  const models: Array<{ alias: string; category?: AudioModelCategory }> = []
+  const models: ModelOption[] = []
   const seen = new Set<string>()
   for (const id of ordered) {
     const channel = channels.find(candidate => candidate.id === id)!
     for (const model of channel.models) {
       if (model.alias === '' || seen.has(model.alias)) continue
       seen.add(model.alias)
-      models.push({ alias: model.alias, ...(model.category === undefined ? {} : { category: model.category }) })
+      models.push({
+        alias: model.alias,
+        ...(model.category === undefined ? {} : { category: model.category }),
+        channelId: channel.id,
+        channelName: channel.name,
+        preset: channel.preset,
+      })
     }
   }
   return models.length > 0 ? { models, defaultChannelId: defaultId } : { models: [], defaultChannelId: defaultId }