Xiaomi MiMo Voice

小米 MiMo V2 TTS 语音合成。支持中文、英文及多种风格（情感、角色扮演、方言、语速控制等）。

Security Scan

The skill is plausibly a TTS client and only needs an API key, but the runtime instructions and the code disagree about where that key is stored (documentation vs actual config lookup), which is an incoherence that could cause misconfiguration and unexpected behavior.

Purpose & Capability

Name/description match the code and instructions: this is a small TTS client that calls an external MiMo TTS API and requires one API key. There are no unrelated credentials or extra binaries requested.

Instruction Scope

SKILL.md tells users to set MIMO_API_KEY under skills.entries in ~/.openclaw/openclaw.json (or export MIMO_API_KEY), but the script's fallback reads ~/.openclaw/openclaw.json at tools.mimoTts.apiKey. The mismatch means following the doc may not provide the key the script actually reads. Otherwise, instructions only ask to install requests and run the script, which is within expected scope.

Install Mechanism

No install spec — instruction-only with a small Python script. Dependency is requests (documented). Nothing is downloaded from untrusted URLs or written to system locations.

Credentials

The only secret needed is an API key, which is proportionate. The script reads MIMO_API_KEY from environment (expected) and also tries to read ~/.openclaw/openclaw.json (expected for local config) but looks under a different JSON path (tools.mimoTts.apiKey) than the SKILL.md recommends (skills.entries.xiaomi-mimo-voice.env). The script only reads that single local config file (read-only), not other credentials or system secrets.

Persistence & Privilege

Skill is not always-enabled, does not request persistent privileges, does not modify other skills or system-wide settings. Autonomous invocation is allowed by default but not combined with other high-risk flags.

What to consider before installing

This appears to be a legitimate TTS script, but the documentation and code disagree about where the API key lives. Before installing or supplying a key: (1) prefer setting MIMO_API_KEY as an environment variable (export MIMO_API_KEY=...) so the script will find it; (2) if you want to put the key in ~/.openclaw/openclaw.json, add it under tools.mimoTts.apiKey or edit the script to read skills.entries.xiaomi-mimo-voice.env.MIMO_API_KEY as the README suggests; (3) inspect the API host (https://api.xiaomimimo.com) to ensure it's the expected service for your key; and (4) if you need stronger assurance, ask the maintainer to fix the documentation/config mismatch or modify the script to accept the documented config path. There is no evidence of data exfiltration, but do not provide high-privilege credentials (AWS, GitHub tokens, etc.) — only the service API key should be used.

Xiaomi MiMo Voice — 语音合成技能

通过小米 MiMo V2 TTS API 将文字转为自然流畅的语音。

前置条件

1. 安装依赖

pip install requests

2. 配置 API Key

编辑 ~/.openclaw/openclaw.json，在 skills.entries 下添加：

{
  "skills": {
    "entries": {
      "xiaomi-mimo-voice": {
        "enabled": true,
        "env": {
          "MIMO_API_KEY": "your-api-key"
        }
      }
    }
  }
}

💡 也可以使用环境变量 export MIMO_API_KEY=xxx，但推荐用 Skill 级别配置，更整洁且仅对该 Skill 生效。

使用方式

python3 <skill_dir>/scripts/tts.py --text "要合成的文字" --output audio.wav

参数

参数	说明	默认值
--text	要合成的文字（必填，可含风格标签）	—
--output	输出音频文件路径	output.wav
--voice	音色	mimo_default
--style	语音风格	—
--format	输出格式：wav / pcm16	wav
--user-text	用户侧消息（辅助上下文）	—

音色

音色	说明
mimo_default	默认音色
default_zh	中文优化
default_en	英文优化

风格列表

类别	风格
情感	Happy · Sad · Angry · Surprised · Fearful · Disgusted · Calm
语速	Speed up · Slow down
角色	Sun Wukong · Lin Daiyu · Zhang Fei · Guan Yu · Zhuge Liang
风格	Whisper · Clamped voice · Taiwanese accent
方言	Northeastern dialect · Sichuan dialect · Cantonese · Henan dialect
其他	唱歌 · Narration · Storytelling

示例

# 基础合成
python3 scripts/tts.py --text "你好，世界！" --output hello.wav

# 情感风格
python3 scripts/tts.py --text "太开心了！" --style Happy --output happy.wav

# 角色扮演
python3 scripts/tts.py --text "俺老孙来也！" --style "Sun Wukong" --output wukong.wav

# 英文
python3 scripts/tts.py --text "Hello world!" --voice default_en --output hello_en.wav

# 方言
python3 scripts/tts.py --text "干啥呢老铁" --style "Northeastern dialect" --output dongbei.wav

# 播报
python3 scripts/tts.py --text "以下是今天的新闻摘要" --style Narration --output news.wav

输出格式

{
  "status": "success",
  "file": "output.wav",
  "size_kb": 128.5,
  "duration_sec": 5.2,
  "voice": "mimo_default",
  "style": "Happy",
  "text": "太开心了！"
}

故障排除

问题	原因	解决方案
未设置 MIMO_API_KEY	未配置 API Key	在 skills.entries.xiaomi-mimo-voice.env 中配置
API 返回异常	Key 无效或过期	检查 Key 是否正确
未获取到音频数据	文本被拒绝	检查文本内容
requests 超时	网络问题	检查网络连接
ImportError: requests	缺少依赖	pip install requests

技术细节

• 模型：mimo-v2-tts
• API：https://api.xiaomimimo.com/v1/chat/completions
• 采样率：24kHz（PCM16 格式）
• 风格标签格式：<style>风格</style> 放在文字开头

更新日志

v1.0.2 (2026-03-24)

• 配置方式统一为 Skill 级别 env（openclaw.json skills.entries）
• 修复依赖说明（openai → requests）
• 文档结构优化

v1.0.1 (2026-03-24)

• 移除 openclaw.json tools 自定义键配置
• API Key 改为环境变量读取

v1.0.0 (2026-03-24)

• 首次发布：中文/英文语音合成，支持情感/角色/方言等多种风格