ShortGPT

ShortGPT is the RayVentura team's open-source English-language short-video automation framework, first released in April 2024, licensed under MIT, and written primarily in Python (★5.5k+ on GitHub). It provides a one-stop ContentShortFactory pipeline: input a topic/script → LLM auto-generates copy → pulls background video/images from Pexels/Unsplash/Pixabay open stock libraries → invokes Edge TTS / ElevenLabs TTS for voiceover → auto-generates subtitles → auto-edits and composites the final short video (9:16 / 1:1 / 16:9). The framework targets English-language YouTube Shorts / TikTok / Instagram Reels automated content production.

Core capabilities center on a "5-stage modular pipeline + LLM-driven creative decision". Stage 1 Script: accepts a custom script, or invokes GPT-4 / Claude / Gemini / Llama to auto-generate (N segments, each with its own prompt). Stage 2 Media: pulls vertical background videos from the Pexels API (free token) + images from Unsplash/Pixabay, automatically aligned to the timeline by segment. Stage 3 Voice: supports Edge TTS (free, no API key) + ElevenLabs (high quality, paid) + OpenAI TTS. Stage 4 Subtitles: auto subtitle alignment + subtitle styling (word-level highlight) + multi-language subtitles. Stage 5 Edit: FFmpeg composites the final video, supports 9:16 (vertical Shorts) / 1:1 / 16:9 + burned-in subtitles + background music overlay + auto B-roll insertion. The overall architecture is loosely coupled — each stage can be replaced independently, making secondary development friendly.

Use cases include: English short-video automation (YouTube Shorts / TikTok matrix accounts), content-farm batch production, automated narration for educational accounts, marketing-account auto-editing, and enterprise overseas social media content. Hardware floor: CPU only; Edge TTS runs in the cloud and needs no GPU; if you choose paid APIs (ElevenLabs / GPT-4 etc.) an API key is required. Onboarding: pip install ShortGPT → configure .env (LLM_API_KEY / PEXELS_API_KEY / TTS choice) → python run_shortgpt.py --topic "AI trends 2026" --vertical 9:16; Colab tutorial is available.

Licensed under MIT, allows commercial closed-source use, modification, and redistribution. Positioning versus VisLane's already-listed money-printer-turbo (Chinese one-stop short-video, Web + API UI, ★95k+): ShortGPT is "English ecosystem + CLI level", best for Python developers integrating into their own pipelines; MoneyPrinterTurbo is "Chinese ecosystem + Web app level", plug-and-play for non-technical users. The two together cover bilingual content matrices — Chinese via MoneyPrinterTurbo, English via ShortGPT. ShortGPT also complements VisLane's pending auto-editor (smart editing) — ShortGPT does content generation, auto-editor does post-editing optimization (removing silence / stutter).