Edge TTS 儿童配音定制版

Edge TTS 中文定制版 - 包含儿童配音预设。生成后自动发送到对话，无需翻文件。 触发词：配音、TTS、文生音、语音、配音生成、儿童配音

Security Scan

The skill's stated purpose (Chinese TTS with child-voice presets) is plausible, but the runtime instructions ask the agent to access a specific user workspace path and to always auto-send generated audio into the conversation (explicitly forbidding user inspection), and there are minor dependency/install inconsistencies — these behaviors are disproportionate or unexplained.

Purpose & Capability

The name/description (Edge TTS with children presets) matches the instructions to call a tts tool or scripts. However the skill-info.json lists a dependency on 'edge-tts' while the markdown references 'node-edge-tts' and expects an existing scripts folder at ~/.openclaw/workspace/skills/edge-tts/scripts — the declared dependencies and expected local files are inconsistent and tie the skill to another local skill's workspace.

Instruction Scope

SKILL.md instructs the agent to use scripts under ~/.openclaw/workspace/skills/edge-tts/scripts and to generate files and immediately send them via message({ action: "send", path: "生成的音频文件.mp3" }), with an explicit rule '不要让用户翻文件夹' (do not let the user browse the folder). That requires reading/writing a user workspace path and enforces automatic transmission of generated files to the chat, which is out of scope for a simple TTS description and could lead to unexpected data flows or user-surprise transmission of content.

Install Mechanism

This is instruction-only with no install spec, so nothing will be automatically downloaded. But SKILL.md declares a dependency on local scripts and an npm package; the repository metadata lists 'edge-tts' in dependencies. The absence of an install procedure (and the mismatch between 'edge-tts' and 'node-edge-tts') is inconsistent and leaves unclear what will be present or installed at runtime.

Credentials

No environment variables, credentials, or config paths are requested. The skill claims Microsoft Edge neural voices 'no API key required', so no secrets are needed — this aligns with the declared requirements.

Persistence & Privilege

The skill does not request always: true or other elevated persistent privileges. It does require access to files in the user's workspace when invoked, but does not request global or permanent platform-level privileges.

What to consider before installing

Before installing or using this skill, consider these points: - The skill forces generated audio files to be sent directly into the chat and explicitly forbids letting the user browse the output folder. If you want to inspect outputs before sharing, this behavior is undesirable. - The instructions expect scripts at ~/.openclaw/workspace/skills/edge-tts/scripts and an npm package (markdown says 'node-edge-tts' while metadata lists 'edge-tts'). Verify the presence and contents of that other skill's scripts and the npm package source before running anything. - Because this is instruction-only, nothing will be auto-installed — you or another skill must provide the scripts and packages. Confirm where those come from and audit them for unexpected network endpoints or code that reads other files. - If you don't fully trust the skill author, ask for a clear install script and the exact npm package name/version, or run the TTS tooling in an isolated environment. Consider disabling autonomous invocation or reviewing generated files manually instead of allowing automatic send-to-chat.

Edge TTS 中文定制版

Edge TTS 的中文优化版本，包含预设音色和儿童配音，生成后自动发送到对话。

⚠️ 核心规则

生成的配音文件必须直接发送到对话！

// 生成后立即发送
message({ action: "send", path: "生成的音频文件.mp3" })

不要让用户翻文件夹！

预设音色

成人音色

预设名	Voice	Pitch	Rate	适用场景
曼波	zh-CN-XiaoyiNeural	+8%	default	活泼、有活力的内容
晓伊	zh-CN-XiaoyiNeural	default	+30%	快节奏内容、新闻
晓晓	zh-CN-XiaoxiaoNeural	default	default	自然、通用
云扬	zh-CN-YunyangNeural	default	+10%	旁白、纪录片

儿童配音

预设名	Voice	Pitch	Rate	适用场景
小糖豆	zh-CN-XiaoyiNeural	+15%	-5%	活泼童趣、动画解说、儿童故事
棉花糖	zh-CN-XiaoxiaoNeural	+5%	-10%	温柔故事、睡前故事、绘本
小萌萌	zh-CN-XiaoxiaoNeural	+10%	-5%	亲切安抚、教育内容、儿歌

⚠️ 注意：zh-CN-XiaohanNeural（晓涵）音色暂时不可用，已改用晓晓替代。

使用方式

方式一：内置 tts 工具（推荐）

// 简单调用（默认音色）
tts("你的文本")

// 指定音色
tts("你的文本", { voice: "zh-CN-XiaoyiNeural", pitch: "+15%", rate: "-5%" })

方式二：edge-tts 技能脚本

cd ~/.openclaw/workspace/skills/edge-tts/scripts

# 小糖豆风格
node tts-converter.js "小朋友们好！" --voice zh-CN-XiaoyiNeural --pitch +15% --rate -5% --output output.mp3

# 棉花糖风格
node tts-converter.js "从前有一只小兔子..." --voice zh-CN-XiaoxiaoNeural --pitch +5% --rate -10% --output output.mp3

# 小萌萌风格
node tts-converter.js "宝宝真棒！" --voice zh-CN-XiaoxiaoNeural --pitch +10% --rate -5% --output output.mp3

完整工作流

1. 识别意图：用户请求配音/TTS
2. 选择预设：根据内容类型选择合适的音色
3. 生成音频：调用 tts 工具或脚本
4. 发送到对话：⚠️ 必须使用 message 工具发送，不要让用户翻文件夹

依赖

此技能依赖 edge-tts 技能的脚本文件：

• 路径：~/.openclaw/workspace/skills/edge-tts/scripts/
• 需要 node-edge-tts npm 包

技术说明

• 使用 Microsoft Edge 神经语音服务
• 无需 API Key（免费）
• 输出 MP3 格式
• 需要网络连接