Install
openclaw skills install video-sttExtract audio from video URLs and transcribe using STT (Speech-to-Text). Supports local Whisper or cloud APIs. Use when: user provides a video URL and wants to know what is being said, transcribing YouTube videos, podcasts, or any video with audio.
openclaw skills install video-stt从视频 URL 提取音频并转换为文字 (Speech-to-Text)
# 进入脚本目录
cd ~/.openclaw/workspace/skills/video-stt/scripts
# 运行转录
bash stt.sh "视频URL"
# 基本用法
bash stt.sh "https://youtube.com/watch?v=xxx"
# 指定输出文件
bash stt.sh "https://youtube.com/watch?v=xxx" -o output.txt
# 使用本地 Whisper 模型
bash stt.sh "https://youtube.com/watch?v=xxx" --local
# 使用云端 API
bash stt.sh "https://youtube.com/watch?v=xxx" --api openai
默认输出纯文本,可选:
.txt - 纯文本.srt - 字幕格式.vtt - WebVTT 字幕.json - 带时间戳的 JSON# OpenAI (如果使用云端)
export OPENAI_API_KEY="sk-xxx"
# 或者使用硅基流动 (更便宜)
export SILICONFLOW_API_KEY="xxx"
# 转录 YouTube 视频
bash stt.sh "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
# 指定模型
bash stt.sh "https://youtube.com/watch?v=xxx" --model medium
# 保存为 SRT
bash stt.sh "https://youtube.com/watch?v=xxx" --format srt
使用 uv 管理 Python 环境:
# 创建虚拟环境
uv venv
uv pip install yt-dlp whisper ffmpeg-python
# 运行
uv run python stt.py "视频URL"