Project Overview
The AI Content Generation Platform is an automated video production system. By feeding the pipeline a reference URL, it extracts raw resources, runs semantic transcript intelligence, writes copywriting scripts, generates voiceovers, synthesizes video assets, and renders a fully composed video with stylized subtitles.
Core Features & Pipeline Workflow
-
Resource Ingestion (RapidAPI): Accepts external source video URLs, utilizing custom RapidAPI endpoints to pull metadata and transcript captions.
-
Semantic Analysis & Copywriting (Gemini API): Dispatches raw content data to Google Gemini API to analyze visual hooks, write engagement-optimized copywriting scripts, and structure scene-by-scene visual instructions.
-
AI Video Generation Lobbies: Routes visual prompts to text-to-video models including Google Veo 3, Kling AI, Luma Dream Machine (Kie AI/Wavespeed) to synthesize fresh, high-fidelity video scenes.
-
Voice Cloning & Audio Generation (ElevenLabs): Takes the written copy script and generates realistic, cloned-voice audio tracks via the ElevenLabs voice generation API.
-
Composition & Rendering (Creatomate): Combines generated video scenes and ElevenLabs voiceovers using the Creatomate API to output the final video asset.
-
Dynamic Captions: Renders multiple subtitle layouts, styling templates, and highlight effects within Creatomate to increase short-form viewer retention.
Technical Architecture
The automation engine coordinates multi-platform APIs. The workflow ingests reference links and downloads captions via **RapidAPI**. **Gemini API** processes these transcripts to compose video copy scripts and detailed visual prompts. Visual frames are generated through external **Veo 3, Kling AI, and Luma** endpoints. Concurrently, **ElevenLabs** renders the voiceover file. The pipeline finishes by packaging these assets into a JSON composition layout sent to the **Creatomate API**, returning the compiled MP4 file.