MiniMax TTS (Text-to-Speech)

⚠️ Requires a MiniMax API Key with TTS/Speech access

• Coding Plan key (sk-cp-xxx) → 不支持 TTS，仅限 M2.7 聊天模型
• 通用 API Key → 请在 MiniMax 开放平台 确认您的订阅是否包含语音合成
• TTS 独立套餐: 访问 MiniMax 语音合成产品

Text-to-speech synthesis using MiniMax's speech synthesis API. Supports 40+ languages, multiple voices, adjustable speed/pitch, and various audio formats.

Setup

1. Configure API Key

openclaw config set skills.entries.minimax-tts.apiKey "sk-your-key"

Or add to openclaw.json skills entries:

{
  "skills": {
    "entries": {
      "minimax-tts": {
        "apiKey": "sk-your-key"
      }
    }
  }
}

2. Dependencies

pip install requests

Architecture

~/.openclaw/workspace/skills/minimax-tts/
├── SKILL.md
├── _meta.json
└── scripts/
    └── minimax_tts.py

Usage

From terminal

# Basic TTS - text to speech
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py "你好，欢迎使用 MiniMax 语音合成"

# With voice selection
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py "Hello world" --voice female-shaonv

# Adjust speed and pitch
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py "快速播报新闻" --speed 1.5 --pitch 2

# Save to file
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py "输出到文件" --output speech.mp3

# List available voices
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py --list-voices

# Multi-segment (audiobook/podcast with multiple voices)
MINIMAX_API_KEY="sk-your-key" python3 scripts/minimax_tts.py --segments segments.json --output combined.mp3

From code

from minimax_tts import generate_speech

# Basic TTS
result = generate_speech(text="你好世界", voice="female-shaonv", output_path="hello.mp3")

# With options
result = generate_speech(
    text="这是一段测试语音",
    model="speech-2.8-hd",
    voice="female-tianmei",
    speed=1.0,
    pitch=0,
    audio_format="mp3",
    output_path="test.mp3"
)

Tool Definition

Name: minimax_tts

Input Schema:

{
  "type": "object",
  "properties": {
    "text": {
      "type": "string",
      "description": "Text to convert to speech (max 10000 chars per request)"
    },
    "model": {
      "type": "string",
      "enum": ["speech-2.8-hd", "speech-2.6-hd", "speech-02-hd"],
      "default": "speech-2.8-hd",
      "description": "TTS model: speech-2.8-hd (recommended, high quality with emotion control), speech-2.6-hd (Coding Plan NOT supported), speech-02-hd (compact HD voice)"
    },
    "voice": {
      "type": "string",
      "default": "female-tianmei",
      "description": "Voice ID: female-tianmei, female-shaonv, female-yujie, female-chengshu, male-qn-qingse, male-qn-jingying, male-qn-badao, male-qn-daxuesheng"
    },
    "speed": {
      "type": "number",
      "default": 1.0,
      "minimum": 0.5,
      "maximum": 2.0,
      "description": "Speech speed, range 0.5-2.0, default 1.0"
    },
    "pitch": {
      "type": "number",
      "default": 0,
      "minimum": -12,
      "maximum": 12,
      "description": "Voice pitch adjustment, range -12 to 12, default 0"
    },
    "audio_format": {
      "type": "string",
      "enum": ["mp3", "wav", "pcm"],
      "default": "mp3",
      "description": "Output audio format"
    },
    "output_path": {
      "type": "string",
      "description": "Save audio to this file path. Default: saved to ~/.openclaw/workspace/tmp/tts_<timestamp>.mp3"
    }
  },
  "required": ["text"]
}

Output: JSON with success status and audio file path

Common Voice IDs

Multi-Segment Generation (Audiobooks / Podcasts)

For multi-character content, create a segments.json file and use --segments:

python3 scripts/minimax_tts.py --segments segments.json --output combined.mp3

segments.json Format

[
  {
    "text": "Morning sunlight streamed into the room.",
    "voice": "female-tianmei",
    "speed": 1.0
  },
  {
    "text": "Welcome to today's episode.",
    "voice": "male-qn-qingse",
    "speed": 1.0
  }
]

Error Codes