DubTok
• Developed DubTok: Architected and implemented DubTok, a Next.js and FastAPI-based web application for automatic video dubbing, leveraging OpenAI Whisper, OpenAI GPT-4, and PlayHT APIs for transcription, translation, and voice cloning, respectively.
• Audio Processing: Spearheaded an audio dubbing pipeline using FFMpeg for video/audio separation, Demucs for vocal isolation, and precise timing adjustments, ensuring natural cadence and flawless synchronization in dubbed videos.
• Deployment: Engineered robust backend infrastructure with FastAPI and SQLite, efficiently managed video storage on AWS S3, and ensured seamless integration of diverse AI and multimedia technologies, optimizing performance for high scalability. Deployed backend on Railways and frontend on vercel