SubtitulosGratis.online: browser-based AI subtitle generation
Web tool to extract audio, transcribe locally, and export editor-compatible SRT subtitles.
Step 01
Problem
Users needed subtitle generation without uploading sensitive media to third-party transcription backends.
Step 02
Context and constraints
Privacy-oriented architecture using FFmpeg WebAssembly and client-side Whisper, deployed on Railway.
Role: Product owner and implementer across frontend architecture, legal/SEO setup, and support operations.
Step 03
Key decisions
- Local browser processing as default model to reduce privacy friction.
- Device-aware execution with WebGPU first and CPU/WASM fallback.
- Bilingual landing and full legal structure for public-facing operation.
Step 04
Outcomes
- Usable upload-to-SRT workflow without account dependency.
- Clear compatibility with SubRip-based editing workflows.
- Solid SEO and legal base for future scaling.
Metrics
- Two execution modes: GPU (WebGPU) and CPU/WASM fallback.
- Direct export of standard `.srt` subtitle files.
- Bilingual ES/EN coverage in main interface and legal pages.
Step 05
Learnings and next improvement
Learnings
- Privacy-preserving UX is a strong product differentiator in creator tools.
- Client-side AI products need explicit hardware expectation management.
Step 06
Visual resources
Stock visuals are active to keep layouts stable. You can replace them later from `src/lib/site.ts` or per-project content.
