AkiFactory is a desktop studio for YouTube Shorts, TikTok and Instagram Reels. You type a topic and it writes the script, records the voiceover, finds or generates the visuals, adds subtitles and edits everything into a ready-to-upload 9:16 video. It's built to be free and local-first, and it works together with H3 Studio on Pinokio for AI-generated footage.
What it does
- Script: writes a hook-first narration with a local LLM through Ollama (e.g. Qwen 3.5), or with an optional cloud model. Includes niche templates such as history & mystery.
- Voiceover: Edge TTS neural voices, plus local engines (XTTS, Chatterbox, F5-TTS, VoxCPM) and RVC voice conversion for your own custom voices.
- Visuals: Pexels / Pixabay stock clips, your own files, or AI clips from H3 Studio. AkiFactory writes the scene prompts, sends them to H3 Studio, chains the shots for a seamless 45–60 s video and checks every clip before editing.
- Subtitles: synced with Whisper; TikTok-style, YouTube-style or minimal presets.
- Editing: FFmpeg montage with background music, intros and outros, and NVENC encoding.
- Platform presets: TikTok, YouTube Shorts, Instagram Reels and LinkedIn.
- Publishing pack: title and description with A/B alternatives, YouTube tags, hashtags and posting-time tips.
- More: batch mode (up to 10 topics), long-form videos with chapters, dual-language TR + EN output, a gallery, statistics and Telegram notifications.
Built for one GPU: steps run one at a time, and Ollama and ComfyUI release VRAM between them, so a single 16 GB card can handle the whole pipeline.
Privacy: API keys stay on your machine and are never uploaded.
Screenshots below 👇



AkiFactory is a desktop studio for YouTube Shorts, TikTok and Instagram Reels. You type a topic and it writes the script, records the voiceover, finds or generates the visuals, adds subtitles and edits everything into a ready-to-upload 9:16 video. It's built to be free and local-first, and it works together with H3 Studio on Pinokio for AI-generated footage.
What it does
Built for one GPU: steps run one at a time, and Ollama and ComfyUI release VRAM between them, so a single 16 GB card can handle the whole pipeline.
Privacy: API keys stay on your machine and are never uploaded.
Screenshots below 👇