Captih is a speech-to-text SaaS built for creators and teams who need accurate transcriptions fast. Upload any video or audio file and receive a synchronized SRT file plus a clean full transcript - no editing, no waiting. Powered by a fine-tuned Whisper model (FARUKxAUTO on HuggingFace), trained on 93 hours of data targeting Tunisian Arabic and accented speech - the best performing open model for this dialect. Stack: Next.js frontend, Tailwind 4, GSAP animations, and a custom inference pipeline. Built to ship real flow with no waitlist gate.