Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/ZJU-REAL/Easelnpx agentmods add skills/zju-real/easel/multi-voice-dubbingWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/zju-real/easel/multi-voice-dubbing)<a href="https://agentmods.dev/skills/zju-real/easel/multi-voice-dubbing"><img src="https://agentmods.dev/badge/skills/zju-real/easel/multi-voice-dubbing/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/zju-real/easel/multi-voice-dubbing"><img src="https://agentmods.dev/badge/skills/zju-real/easel/multi-voice-dubbing.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Data Exfiltration · line 34 Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.Fix: Verify the destination URL is trusted and necessary. Remove or replace with documented APIs. Ensure no secrets, tokens, or PII are transmitted.
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00111 | $0.02339 |
| Opus 5 | $0.00056 | $0.01170 |
| Sonnet 5 | $0.00022 | $0.00468 |
| Haiku 4.5 | $0.00011 | $0.00234 |
Grade A, and why
multi-voice-dubbing scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
多角色 / 对话配音（Multi-voice Dubbing）
把「多人对话 / 多角色脚本」合成为多声线音轨——每个角色一个符合其人设的声音，
不再是「全程一个声音」。核心引擎 skills/shared/scripts/multivoice.py：
逐行调 tts.py（edge-tts 免费多音色）或 voice_clone.py（云端表现力 provider / 克隆音色）合成，
把每行 emotion 喂进 provider 的真情感通道，ffmpeg 拼成一轨 + 生成对齐的说话人字幕。
创意（谁说什么、什么情绪）由你 LLM 产出；音色映射在 cast。确定性 IO（逐行合成/拼接/字幕）走引擎。 产物
voice.mp3可直接当 narration 喂auto-short-video/assemble.py或加进任意视频；voice.srt是带角色名的字幕。
配音质量分层（重要：治「像 AI 平读」）
edge-tts 没有情感引擎，只能变速变调，再怎么调也像机器平读。要"像人"必须用有情感通道的云 provider（用户自备 key，无需 GPU）：
引擎（cast 里 engine） |
质量 | 需要 | 情感机制 |
|---|---|---|---|
edge（默认，免费兜底） |
⚠️ 平、机器感，仅供草稿 | 无 key、外网 | 仅 rate/pitch/volume 微调 |
clone+openai-compatible→SiliconFlow CosyVoice2（推荐） |
好、中文自然 | VOICE_API_KEY（便宜/新用户赠额） | 内联 <|endofprompt|> 指令 |
clone+gemini |
好、真免费 | GEMINI_API_KEY（Google 免费层，国内需代理） | 自然语言前缀 |
clone+minimax / dashscope |
好、有情绪 | 各家 key | emotion 枚举 / instruct |
推荐 SiliconFlow（云端 CosyVoice2、无需 GPU、中文最稳）配置落 .env：
VOICE_PROVIDER=openai-compatible
VOICE_BASE_URL=https://api.siliconflow.cn/v1
VOICE_API_KEY=<你的key>
VOICE_MODEL=FunAudioLLM/CosyVoice2-0.5B
VOICE_INSTRUCT_MODE=inline
cast 里角色：--engine clone --provider openai-compatible --voice-id FunAudioLLM/CosyVoice2-0.5B:alex（8 音色 alex/anna/benjamin/…）。
- 逐行 emotion 自动驱动演绎：
lines.json每行的emotion（愤怒/崩溃大哭/冷笑/温柔…）→ 引擎按 provider 转成对应情感参数。写具体越贴戏越好。 - 想要「像人」→ 至少给主角/关键角色配一个云 provider（
engine=clone）；配了 key 才有情绪，没 key 自动回退 edge（平）并告警。 - provider 配置见
voice_clone.py头部（各家 env）；voice_clone.py check --provider <名>离线校验 key 是否齐。
谁会用到
短剧对白（short-drama 已内部委派）、论文双人问答讲解（paper-explainer：主讲+提问者）、 访谈/播客脚本、有声剧、任何「多个说话人」的口播。单人整段口播用 tts-voiceover 即可。
输入
| 字段 | 必填 | 说明 |
|---|---|---|
| cast.json | 是 | 选角表：每个说话人 → 音色（edge 音色 + pitch/rate，或克隆音色）。含「旁白/主讲」条目 |
| lines.json | 是 | 逐行对白：有序 [{speaker, text, emotion}]（speaker 用 cast 里的名字；emotion 如 冷/怒/紧张/温柔，自动匹配语气） |
| 输出路径 | 否 | voice.mp3（默认与调用方约定）；voice.srt 同名 |
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 129 lines · 111 tokens per session scan A e64e4b9e78d9
multi-voice-dubbing is a skill published in the GitHub repository ZJU-REAL/Easel (841 stars, last pushed yesterday), licensed Apache-2.0. It adds 111 tokens to every session and 2,339 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
python-run
Run and debug Python scripts in the project. Use when the user says "run python", "execute this script", "debug this py file", or wants to run/modify a .py file. Handles dependency checks, linting, execution, and error analysis.
roadtrip-navigator
Generate North American road-trip itineraries as a map-first, offline-friendly single-file HTML page. Plans around daily driving segments, overnight stops, fuel/EV-charging, national-park reservations (Recreation.gov / NPS), seasonal road closures, and timezone/border crossings — for executable, decision-ready trips.…
skill-authoring
Author SKILL.md: frontmatter, structure, writing principles.
systematic-debugging
4-phase root cause debugging: understand bugs before fixing.
github-code-review
Review PRs: diffs, inline comments via gh or REST.
simplify-code
Sequential 3-lens cleanup of recent code changes.