shadow-shinigami/lyric-video-director — explained in plain English
Analysis updated 2026-05-18
Generate a typography led lyric video from a song and its exact lyric timings
Automatically check that on screen lyric text matches the real lyrics exactly
Create a wordless animated narrative for instrumental sections of a song
Assemble final video with FFmpeg while preserving the original master audio
| shadow-shinigami/lyric-video-director | alibaba/omnidoc-tokenbench | arccalc/dwmfix | |
|---|---|---|---|
| Stars | 43 | 43 | 43 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | moderate | easy |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | researcher | general |
Figures from each repo's GitHub metadata at analysis time.
Requires access to GPT Image 2 and Gemini Omni Flash generation tools plus a local ffmpeg install.
Lyric Video Director is a skill for Hermes Agent, an AI agent framework, that automates the process of turning a song, its exact lyrics, and rough timing of each line into a stylized lyric video where the words themselves are the visual star of the show. The workflow treats the lyric text as the main character on screen. First, an AI image tool called GPT Image 2 designs typography focused compositions, meaning scenes built around how the words look and move rather than live action footage. Then a video tool called Gemini Omni Flash takes those still designs and animates them into motion. Throughout the process, automated checks make sure the text that appears on screen exactly matches the real lyrics, and that the original song audio is never altered or replaced. Before committing to full video generation, the skill builds a visual style guide and short test sequences, since generating final videos in bulk is expensive and mistakes are costly to redo. It can also handle instrumental sections with no lyrics by creating a wordless visual narrative that still flows naturally from one scene to the next. Once everything is generated, FFmpeg, a well known open source video and audio tool, assembles the final video and layers it back over the original, untouched master recording of the song. A final round of checks covers technical quality, whether the on screen text is accurate, whether it matches the music, and whether the creative style stays consistent throughout. To use it, you copy or clone the project into a specific skills folder inside your Hermes Agent profile, after which Hermes can discover and use it automatically. It expects Python 3, the ffmpeg and ffprobe command line tools, and access to the GPT Image 2 and Gemini Omni Flash generation tools, with an optional extra tool for deeper audio analysis. It is written in Python and released under the MIT license.
A Hermes Agent skill that turns a song and its exact lyrics into a stylized, typography focused lyric video, with automatic checks that the on screen text and audio stay accurate.
Mainly Python. The stack also includes Python, FFmpeg, GPT Image 2.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
double-check against the repo, no cap.