Feishu Voice Bubble

Send a native Feishu voice message (playable bubble, not attachment) using MiniMax TTS.

Setup

1. Configure voice_config.json

Create ~/.openclaw/workspace/skills/minimax-feishu-voice/voice_config.json:

{
  "voice_id": "female-yujie-jingpin",
  "speed": 1.0,
  "vol": 1.0,
  "pitch": 0,
  "api_key": "YOUR_MINIMAX_API_KEY"
}

• api_key: 你的 MiniMax TTS API Key（必填）
• voice_id: 语音ID（见下方列表）
• speed/vol/pitch: 可选参数

2. Recommended Voice IDs

完整列表：https://platform.minimaxi.com/docs/faq/system-voice-id

默认值：female-yujie-jingpin

Quick Use

python3 ~/.openclaw/workspace/skills/minimax-feishu-voice/scripts/send_feishu_voice.py "<text>" <飞书用户的open_id>

Example

python3 ~/.openclaw/workspace/skills/minimax-feishu-voice/scripts/send_feishu_voice.py \
  "你好，这是一条语音消息" \
  "<飞书用户的open_id>"

How It Works

1. Reads api_key from voice_config.json
2. Calls MiniMax TTS API → hex audio → MP3
3. Decodes MP3 → PCM → re-encodes as opus in OGG container at 16kHz (Feishu requirement)
4. Extracts duration_ms from WAV header
5. Uploads to Feishu with file_type=opus + duration param → gets file_key
6. Sends msg_type: "audio" message with file_key → native voice bubble

Prerequisites

• MiniMax TTS API Key in voice_config.json → api_key
• Feishu channel configured with appId + appSecret in openclaw.json
• ffmpeg installed with libopus support

Notes

• Feishu requires 16kHz sample rate (not 32kHz or 24kHz)
• Duration (ms) must be included in upload request for bubble display
• User's open_id is available as sender_id in message metadata