Caption CC
Code-Switching Subtitle Generator
Caption CC generates accurate English subtitles from Tamil-English code-switched video, with optional AI dubbing. A 5-stage caption pipeline and 7-stage dubbing pipeline handle everything from download to export, supporting YouTube URLs and direct file uploads up to 500 MB.
5
Caption Stages
7
Dubbing Stages
10 min
Max Duration
SRT, VTT, MP4
Export Formats
Caption Pipeline
Five-stage pipeline: download or upload → FFmpeg extracts WAV (16kHz mono) → Groq Whisper transcribes → Gemini corrects code-switching and translates to English → subtitle generation as SRT, VTT, or burned-in MP4. Subtitle segments are editable in-app before export. Real-time progress via Server-Sent Events.
AI Dubbing Pipeline
Seven-stage dubbing flow adds TTS and audio assembly on top of captioning: after translation, ElevenLabs synthesizes each segment, FFmpeg assembles the full audio track, and the dubbed audio is merged back into the video. The final dubbed MP4 is available for one-click download. Timing is best-effort matched to original segment boundaries.
Architecture & Limits
Full-stack TypeScript monorepo with npm workspaces (client + server). MongoDB tracks job state, subtitle segments, and usage per user. LLM temperature set to 0.3 with 3 retries and backoff. Supports MP4, MP3, WAV, M4A, MOV, WEBM. Max file size 500 MB, max duration 10 minutes.
Operational Value
Cuts manual captioning and translation effort. Expands regional content into English-speaking audiences. Creates reusable subtitle and dubbed media assets for publishing.
Tech Stack
Full Stack
Related Service