What It Is
Give 4ort-tv a topic and it hands back a finished long-form MP4: no editing suite, no human in the middle. It's a single end-to-end pipeline: an LLM script writer drafts the piece, AI slide generation (Presenton) builds the visuals, neural TTS (Kokoro) narrates it, and original AI music (ACE-Step) scores it.
The final stage is GPU composition via HyperFrames, where script, slides, narration, and music get assembled into the finished video on an RTX 4090. Every layer of the production (words, visuals, voice, soundtrack) is generated, then composed into one continuous render.
It's the long-form sibling of 4ort.stream's shorts factory: where that pipeline cranks out autonomous shorts overnight, 4ort-tv takes the same topic-in, video-out philosophy and stretches it to full-length productions, driven from the command line.
Highlights
- Topic in, MP4 out: a single end-to-end pipeline from prompt to finished long-form video
- LLM script writer: the entire script is drafted by the pipeline, not a person
- AI slide generation: visuals built automatically with Presenton
- Neural narration: Kokoro TTS voices every script
- Original AI score: ACE-Step generates the music; no stock library
- GPU composition: HyperFrames assembles the final render on a CUDA RTX 4090
Status
● Working v1 pipeline (CLI).