- TeX 60.2%
- Python 20.6%
- HTML 7.1%
- Shell 6.4%
- JavaScript 3.9%
- Other 1.8%
| Filename | Latest commit message | Latest commit date |
|---|---|---|
| bin | ||
| dashboard | ||
| protocol | ||
| state | ||
| visualizer | ||
| academize.sh | ||
| autopush.sh | ||
| build_manifest.py | ||
| common_io.py | ||
| config.py | ||
| discover.sh | ||
| extract-text.py | ||
| get-context.sh | ||
| LICENSE | ||
| Protocol Suite.pdf | ||
| Protocol-Suite.tex | ||
| protocol_suite.zip | ||
| publish.py | ||
| README.md | ||
| render.py | ||
| requirements.txt | ||
| schedule.py | ||
| select.py | ||
| walk.sh | ||
Cakewalk
UNTESTED
Protocol Suite
Content engine
Cakewalk is an autonomous system that discovers source material, transforms it into video, and schedules it for publication over time.
It is designed to run continuously, converting a living corpus into a stream of videos with minimal intervention.
What it does
Cakewalk operates as a pipeline:
- Scans repositories for audio files with matching subtitles
- Builds a candidate set of publishable media
- Generates video outputs using ffmpeg and visualizers
- Attaches metadata and descriptions
- Schedules videos across future timestamps
- Publishes automatically (or simulates publishing)
It can run once, or indefinitely via cron.
Core idea
Cakewalk treats content as a continuous process rather than a one-time task.
Each run advances the system forward:
source material → candidates → rendered videos → scheduled outputs
Over time, this produces a structured, time-ordered channel from an evolving corpus.
Project structure
cakewalk/
walk.sh # main pipeline entrypoint
discover.sh # finds MP3 + VTT pairs
build_manifest.py # creates candidate objects
select.py # chooses items to process
render.py # builds videos with ffmpeg
schedule.py # assigns publish times
publish.py # uploads or simulates upload
candidates/ # discovered media
configs/ # per-video configuration
build/ # rendered videos
schedule/ # scheduled publish entries
state/ # persistent system state
dashboard/ # browser-based control surface
visualizer/ # rendering components
Quick start
chmod +x walk.sh discover.sh
./walk.sh
This will:
- discover content
- generate configs
- render videos
- schedule them
- simulate publishing
Running continuously
Set up a cron job:
crontab -e
0 2 * * * /path/to/cakewalk/walk.sh >> log.txt 2>&1
Cakewalk will now run daily and extend the schedule over time.
Dashboard
Start the local dashboard:
python3 dashboard/server.py
Open:
http://localhost:8000
The dashboard allows:
- previewing candidate videos
- editing titles, descriptions, and parameters
- selecting visualizers
- preparing items before scheduling
Rendering
Videos are generated using ffmpeg.
Current features include:
- subtitle burn-in
- audio playback control (speed, tone-ready)
- pluggable visualizers (Lissajous, waveform, etc.)
The visualizer system can be extended independently.
Publishing
publish.py currently simulates uploads.
To enable real uploads:
- configure YouTube Data API credentials
- replace the mock uploader with API calls
Description generation
Cakewalk can attach structured descriptions using:
- extracted text (PDF / EPUB / MHTML)
- chunked summaries
- repository metadata
This allows descriptions to be generated automatically from source material.
Philosophy
Cakewalk is not just an uploader.
It is a system that:
- treats content as a continuous process
- separates discovery, transformation, and publication
- allows both autonomous operation and manual intervention
Over time, it becomes a persistent engine that converts structure into media.