MiniMax TTS for FeiShu

MiniMax 文字转语音，支持中文音色、自动情绪检测、语气词音效和停顿标记

Security Scan

The skill's code, instructions, and requested credentials are coherent with a MiniMax TTS → Feishu integration; no unexplained credential or network access is present, though there are minor implementation bugs to be aware of.

Purpose & Capability

The skill claims to call MiniMax TTS and send audio to Feishu and the code indeed calls api.minimaxi.com and open.feishu.cn and requires MINIMAX_API_KEY, FEISHU_APP_ID, and FEISHU_APP_SECRET. No unrelated credentials or binaries are requested. The files and environment variables align with the stated purpose.

Instruction Scope

Runtime instructions and SKILL.md match the code paths (tts, design, list, update, save/trigger). The skill reads/writes local cache and voice-map files (/tmp/last_miss_m_message.txt, voices-map.md) and calls Feishu message APIs to fetch and send messages — these are expected for this integration. Note: there are implementation bugs that may cause runtime failures (index.js spawns python3 on a shell script; voice_design.py calls send_audio_message without importing/defining it) which are correctness issues but not evidence of hidden/excessive data access.

Install Mechanism

No install spec is provided (instruction-only install), and included code is plain Python/JS/bash that will run in the environment. No remote download URLs, archive extraction, or surprising installer behavior detected.

Credentials

Requested environment variables (MINIMAX_API_KEY, FEISHU_APP_ID, FEISHU_APP_SECRET, optional FEISHU_USER_OPEN_ID, and optional TTS_VOICES_MAP_PATH) are proportional and necessary for TTS generation and Feishu integration. No unrelated secrets or extra third-party tokens are requested.

Persistence & Privilege

The skill does not request always:true and does not modify other skills or system-wide agent settings. It writes local files (voice map and /tmp caches) within expected scope. Autonomous invocation is allowed (platform default) but not combined with elevated privileges.

Assessment

This skill appears to do what it says: convert text using MiniMax and send audio to Feishu. Before installing, ensure you: 1) provide only scoped MiniMax and Feishu credentials and avoid committing them; 2) restrict FEISHU_USER_OPEN_ID to the intended recipient(s); 3) run the skill in a non-production/sandbox environment first — there are implementation bugs (index.js invokes python3 on a shell script, and voice_design.py references send_audio_message without importing it) that can cause errors; 4) verify network policy so only api.minimaxi.com and open.feishu.cn are contacted; and 5) if you need stronger assurance, request the maintainer fix the noted bugs and review the code that fetches message content (it will read messages via Feishu API when given valid app credentials).

minimax-tts

MiniMax 文字转语音 Skill，支持中文音色选择、情绪检测、停顿标记。

核心功能

1. 文字转语音 (tts)

生成语音并发送到飞书，支持自动情绪检测和停顿标记。

tts_wrapper.sh tts "<文本>" [voice_id] [user_open_id]

2. 音色设计 (voice_design)

通过描述创建自定义音色。

tts_wrapper.sh design "<音色描述>" "<试听文本>" "<要说的内容>"

3. 查询音色列表 (list)

tts_wrapper.sh list

4. 更新音色目录 (update)

从 API 获取最新音色并更新本地 voices-map.md。

tts_wrapper.sh update

---

交互式语音生成

使用方式

在飞书对话中输入 "转语音" 即可触发。

• 直接发送：转语音 → 将我最近一条消息转成语音
• 回复某条消息发送：转语音 → 将那条消息转成语音

支持的触发词

• 转语音
• 转成语音
• 变成语音
• 说一遍
• 语音播放

工作原理

1. 用户说"转语音"后，系统从以下来源获取要转换的文字：
   • 被回复的那条消息（如果有）
   • 我最近发送的文字（保存在本地）
2. 对文字进行预处理（情绪检测 + 停顿标记）
3. 调用 MiniMax TTS API 生成语音
4. 将语音消息发送到飞书

---

文本预处理功能

自动情绪检测

系统会根据文本关键词自动判断情绪，并在 API 参数中使用：

情绪	关键词示例
happy	开心、高兴、太好了、哈哈、太棒了
sad	伤心、难过、遗憾、可惜、唉
angry	生气、气死了、讨厌、哼
fearful	害怕、担心、恐怖
disgusted	恶心、厌恶
surprised	惊讶、吃惊、没想到
calm	平静、淡定、好吧

停顿标记

文本会自动在标点后插入停顿标记，提升语音自然度：

• ， → 0.3秒停顿
• 。 → 0.5秒停顿
• ？ → 0.5秒停顿
• ！ → 0.5秒停顿

示例：

• 原文：真的吗？太好了！哈哈哈！
• 处理后：真的吗？<#0.5#>太好了！<#0.5#>哈哈哈！<#0.5#>
• 情绪：happy

手动指定情绪

如需强制指定情绪，可传入 emotion 参数覆盖自动检测。

---

环境变量

变量	说明
MINIMAX_API_KEY	MiniMax API Key
FEISHU_APP_ID	飞书 App ID
FEISHU_APP_SECRET	飞书 App Secret

---

文件结构

skills/minimax-tts/
├── SKILL.md
├── index.js
└── scripts/
    ├── tts.py                  # 核心 TTS 功能
    ├── voice_design.py         # 音色设计
    ├── list_voices.py          # 查询音色列表
    ├── update_voices_map.py    # 更新音色目录
    ├── text_preprocessor.py   # 文本预处理（情绪+停顿）
    ├── tts_from_chat.py        # 对话式转语音
    └── tts_wrapper.sh          # 统一入口

默认音色

Chinese (Mandarin)_Gentle_Senior（温柔学姐） - 温柔、知性、略带亲切感的声音