Show HN: We built a subtitle generator that auto-detects real speaker names

Hi HN! We built a subtitle generator that solves some real pain points we encountered while working with video content.

*The problem:* Most subtitle tools give you "Speaker 1, Speaker 2, Speaker 3" labels, require manual cleanup of filler words, and have no way to handle sensitive content that shouldn't be exported.

*What we built:*

- *AI Speaker Detection*: Automatically suggests real names from self-introductions ("Hi, I'm John" → suggests "John"). Works for both English and Chinese.

- *Bilingual Filler Removal*: One-click removal of 35+ filler words (um, uh, like, 嗯, 这个, 那个). Most tools only support English – we handle both English and Chinese.

- *Off-Record Protection*: Mark sensitive segments once, they're automatically excluded from all exports. Essential for journalists and legal professionals.

*Tech stack:* Next.js, TypeScript, Whisper AI for transcription, custom NLP for speaker detection and filler word identification.

*Free tier:* 30 minutes/month, no signup required, no watermarks. Export clean SRT/VTT files immediately.

Try it at: https://harku.com/tools/subtitle-generator

We're especially proud of the Chinese language support – it's surprisingly hard to find good subtitle tools that handle Chinese filler words and name detection properly.

Would love to hear your feedback, especially if you work with multilingual content or handle sensitive information that needs redaction.