| name | prismtek-youtube-buddy-lipsync |
| description | Produce a Prismtek/Buddy talking-host overlay for narrated videos using pixel-art visemes and FFmpeg audio RMS animation. |
| version | 1.1.0 |
| author | Prismtek |
| license | MIT |
| platforms | ["macos","linux"] |
| metadata | {"tags":["youtube","video","ffmpeg","imagemagick","pixel-art","vtuber","buddy","prismtek"],"canonical_runtime":"codysumpter-cloud/buddy-agent#18"} |
Prismtek YouTube Buddy Lip-Sync
This skill gives Omni Buddy a local-device reference for rendering a Buddy talking-host overlay from a clean Buddy avatar and narrated media.
The canonical runnable package lives in buddy-agent. This repo keeps the embodied/local-device integration reference so the Raspberry Pi / offline Buddy surface can use the same contract without inventing a separate workflow.
Expected Local Tools
command -v magick
command -v ffmpeg
command -v ffprobe
python3 --version
Flow
- Generate Buddy mouth states from a clean Buddy avatar.
- Analyze narration energy frame-by-frame.
- Render a proof clip with a Buddy overlay.
- Review visual quality before using the output.
Generated Outputs
Keep generated media and receipts outside the repo, preferably under a local output folder.
Related
- Canonical runtime PR:
codysumpter-cloud/buddy-agent#18
- Operator reference PR:
codysumpter-cloud/buddy-brain#315
- Upstream context:
NousResearch/hermes-agent#26463
