MiniMax TTS Plus

Multi-agent + multi-channel native voice message TTS skill.

Core Script

All operations go through tts-xiaoye.sh (TTS generation + channel delivery).

Quick Start

bash tts-xiaoye.sh "Text to speak"

Multi-Channel Usage

Send Feishu Native Voice Message (Full Flow)

OPUS=$(bash tts-xiaoye.sh --feishu "Feishu voice content" 2>/dev/null | python3 -c "import sys,json; print(json.load(sys.stdin)['audio_file'])")
openclaw message send --channel feishu --account <YOUR_ACCOUNT_ID> --target <FeishuUserID> --media "$OPUS"

Multi-Agent Voice Configuration

Each agent can choose a unique voice and write it into their SOUL.md Voice Identity section:

## Voice Identity
- TTS model: speech-2.8-hd
- TTS voice: Chinese (Mandarin)_Warm_Girl
- TTS script: scripts/tts-xiaoye.sh

Recommended voices (verified):

List Available Voices

python3 tts-xiaoye.sh --list-voices
# or directly:
python3 scripts/tts.py --list-voices

This calls the MiniMax API and prints all available voices organized by category (System Voices, Cloned Voices, Generated Voices).

Available Models

Full Parameters

tts-xiaoye.sh --text "Text" [--voice VoiceID] [--model Model] [--caption Caption]

Technical Notes

• TTS outputs MP3 natively. Telegram sends directly via Bot API sendVoice (MP3 supported natively).
• Feishu native voice messages require OGG/Opus format. FFmpeg handles conversion (~25ms per audio, negligible).
• FFmpeg installation: brew install ffmpeg (Linuxbrew/macOS) or apt install ffmpeg (Linux).

Setup

1. Copy setup.txt to .env and fill in your credentials:

cp skills/voice-for-openclaw/setup.txt skills/voice-for-openclaw/.env
# Then edit .env with your real values

1. The script loads credentials from .env at runtime — no hardcoded tokens in scripts.

Required env vars:

> ⚠️ Security note: Credentials are loaded from .env only — no tokens are hardcoded in shell scripts. The .env file is gitignored and never published.

> 💡 API endpoint: The TTS API uses https://api.minimaxi.com (MiniMax's official API server), which is separate from the developer portal at platform.minimax.io.

MiniMax TTS Plus（多语言增强版）

多 Agent + 多渠道原生语音条增强版 TTS 技能。

核心脚本

所有操作通过 tts-xiaoye.sh 完成（TTS 生成 + 渠道发送）。

快速使用

bash tts-xiaoye.sh "要转语音的文字"

多渠道用法

发送飞书原生语音条（完整流程）

OPUS=$(bash tts-xiaoye.sh --feishu "飞书语音内容" 2>/dev/null | python3 -c "import sys,json; print(json.load(sys.stdin)['audio_file'])")
openclaw message send --channel feishu --account <YOUR_ACCOUNT_ID> --target <飞书用户ID> --media "$OPUS"

多 Agent 音色配置

每个 Agent 可以选择不同音色，写入各自的 SOUL.md 的 Voice Identity 节即可：

## Voice Identity
- TTS model: speech-2.8-hd
- TTS voice: Chinese (Mandarin)_Warm_Girl
- TTS script: scripts/tts-xiaoye.sh

推荐音色（已验证）：

可用模型

完整参数

tts-xiaoye.sh --text "文字" [--voice 音色ID] [--model 模型] [--caption 文字]

技术说明

• TTS 原生输出 MP3，Telegram 直接发送（Bot API sendVoice 支持 MP3）
• 飞书原生语音条需要 OGG/Opus 格式，通过 FFmpeg 转换（耗时约 25ms/音频，可忽略）
• FFmpeg 安装方式：
• macOS/Linuxbrew: brew install ffmpeg
• Ubuntu/Debian: apt install ffmpeg