VoxCPM中文配音

🎯 **唯一使用VoxCPM的中文配音技能** - 外语视频一键中文配音，支持硬字幕检测、断点续传、智能BGM。触发场景：(1) 用户需要给外语视频配音 (2) 视频翻译需求 (3) 多语言内容本地化

This package implements a plausible VoxCPM-based dubbing pipeline, but inspect and/or remove the uploader before running. Specific actions to consider: - Review scripts/upload_bilibili.py: it reads credentials from a hard-coded path (D:/openclaw_workspace/credentials/bilibili.json) and will attempt to upload videos. If you don't want uploads or don't trust that behavior, delete or sandbox this file. - Confirm which API endpoints you will use for translation/vision (the defaults point to a third party, SiliconFlow). Only provide API keys that you trust to those endpoints; the code will send base64-encoded frames to the vision endpoint. - Fix the metadata mismatch: registry lists no required env vars but SKILL.md and code expect TRANSLATE_API_KEY and VOXCPM_DIR — ensure you supply only necessary secrets. - Run the tool in an isolated environment (container or VM) and avoid placing sensitive credentials at the hard-coded path. Replace hard-coded paths with explicit, documented config or environment variables. - If you plan to publish/upload, audit the uploader and any third-party libraries (bilibili_api) and adjust paths (ImageMagick/ffmpeg) to match your system. If you want, I can point to the exact lines that read the hard-coded credential path and suggest a safe patch to remove or parameterize the uploader.

Type: OpenClaw Skill Name: video-dubbing Version: 1.0.0 The skill bundle provides a functional video dubbing workflow using Whisper for transcription, VoxCPM for TTS, and SiliconFlow APIs for translation and vision-based subtitle detection. The Python scripts (dubbing.py, add_bgm.py, upload_bilibili.py) utilize standard libraries and subprocess calls to ffmpeg for media processing, which is consistent with the stated purpose. While upload_bilibili.py contains hardcoded Windows-specific paths (e.g., D:/openclaw_workspace/credentials/bilibili.json and E:/ImageMagick.../ffmpeg.exe) that indicate a localized development environment, there is no evidence of malicious intent, data exfiltration, or unauthorized remote control.