audio-generation

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Audio Generation

音频生成

Use When

使用场景

The desired final asset is generated audio or audio prepared for a video render.
The request includes TTS, voice design, voice cloning, voice change, translated dub, podcast audio, or lip-sync handoff.
The next decision is audio task class, reference policy, and runner handoff.

所需最终资产为生成的音频或为视频渲染准备的音频。
请求包含TTS、语音设计、语音克隆、变声、翻译配音、播客音频或唇形同步移交。
下一步需要确定音频任务类别、参考规则以及执行器移交方案。

Do Not Use When

禁用场景

The user only needs transcription, subtitles, or audio analysis. Use
```
media-router
```
.
The voice request is already normalized for execution. Use
```
voice-batch-runner
```
.
The final work is a full video production pipeline. Use
```
video-generation
```
or
```
ugc-flow
```
after the audio handoff is clear.

用户仅需要转录、字幕或音频分析。请使用
```
media-router
```
。
语音请求已标准化可直接执行。请使用
```
voice-batch-runner
```
。
最终工作为完整视频制作流水线。在明确音频移交方案后，请使用
```
video-generation
```
或
```
ugc-flow
```
。

Core Boundary

核心边界

This is the audio generation controller. It does not submit jobs.

It must classify the task and hand off execution. It must not let a runner invent voice strategy, translation policy, or lip-sync intent.

本技能为音频生成控制器，不提交任务。

它必须对任务进行分类并移交执行权，不得让执行器自行制定语音策略、翻译规则或唇形同步意图。

Task Classes

任务类别

Task class	Use when	Handoff
`tts`	new spoken audio from script	`voice-batch-runner` with voice design rules
`change_voice`	preserve script, alter voice identity or delivery	reference contract, then `voice-batch-runner`
`translate_dub`	translate and dub source audio	require language, meaning-preservation, and timing policy
`voice_clone_take`	approved reference voice should preserve timbre	bind reference audio, then `voice-batch-runner`
`podcast_audio`	speaker-led or conversational audio	create voice/script handoff before video assembly
`lip_sync_handoff`	audio drives talking-head or UGC render	`voice-batch-runner` , then `video-generation`

任务类别	使用场景	移交目标
`tts`	根据脚本生成新的语音音频	附带语音设计规则的 `voice-batch-runner`
`change_voice`	保留脚本内容，修改语音身份或表达方式	参照合约后移交至 `voice-batch-runner`
`translate_dub`	对源音频进行翻译并配音	需明确语言、语义保留规则和时长同步规则
`voice_clone_take`	需保留已获批准的参考语音音色	绑定参考音频后移交至 `voice-batch-runner`
`podcast_audio`	主播主导或对话式音频	在视频组装前完成语音/脚本移交
`lip_sync_handoff`	音频驱动虚拟主播或UGC渲染	先移交至 `voice-batch-runner` ，再移交至 `video-generation`

Reference Rules

参考规则

Approved voice reference audio is
```
binding
```
.
Accent, energy, cadence, or genre examples are inspiration-only unless the user explicitly binds them.
Source audio used only for translation meaning is not a voice identity binding unless stated.
Excluded voices, music, or effects must not enter the runner request.

已获批准的语音参考音频具有
```
binding
```
（约束力）。
口音、语气、节奏或风格示例仅作为灵感参考，除非用户明确将其设为约束条件。
仅用于翻译语义的源音频，除非特别说明，不视为语音身份的约束条件。
禁用的语音、音乐或音效不得纳入执行器请求。

Routing Table

路由表

If not audio-generation	Send to
Transcribe or analyze existing audio	`media-router`
Need generated image/video around audio	`video-generation`
Need normalized hosted voice execution	`voice-batch-runner`
Need lip-sync video after audio	`video-generation`

非音频生成场景	移交目标
转录或分析现有音频	`media-router`
需要围绕音频生成图像/视频	`video-generation`
需要标准化的托管语音执行	`voice-batch-runner`
音频生成后需要唇形同步视频	`video-generation`

Output Shape

输出格式

Return:

```
taskClass
```
```
scriptPolicy
```
```
voicePolicy
```
```
referencePolicy
```
```
runnerHandoff
```
```
nextVideoHandoff
```
when lip-sync or video assembly follows
```
mustNotDo
```

返回以下内容：

```
taskClass
```
（任务类别）
```
scriptPolicy
```
（脚本规则）
```
voicePolicy
```
（语音规则）
```
referencePolicy
```
（参考规则）
```
runnerHandoff
```
（执行器移交信息）
若后续涉及唇形同步或视频组装，需返回
```
nextVideoHandoff
```
（下一阶段视频移交信息）
```
mustNotDo
```
（禁止操作）

Fail Fast

快速失败规则

Stop if script, target language, voice identity, reference ownership, timing, or lip-sync target is missing and would change the result.
Do not submit, poll, upload, or call provider-specific routes.
Do not ask
```
voice-batch-runner
```
to decide the creative role of the voice.

若脚本、目标语言、语音身份、参考音频归属、时长同步规则或唇形同步目标缺失，且这些信息会影响结果，则停止流程。
不得提交、轮询、上传或调用特定服务商的路由。
不得让
```
voice-batch-runner
```
决定语音的创意角色。

Public Command Boundary

公开命令边界

Check readiness first:

postplus doctor --skill audio-generation

This public skill is instruction-driven. Produce the controller handoff artifact directly from the available evidence.
Do not call private provider/runtime paths or unpublished local tools.
If the CLI returns a quote-confirmation challenge, run
```
postplus quote confirm --json --challenge-file <challenge.json>
```
and retry with the returned token.

首先检查就绪状态：

postplus doctor --skill audio-generation

。

本公开技能为指令驱动型，需直接根据现有信息生成控制器移交产物。
不得调用私有服务商/运行时路径或未发布的本地工具。
若CLI返回报价确认挑战，请运行
```
postplus quote confirm --json --challenge-file <challenge.json>
```
并使用返回的令牌重试。