We just solved real-time translation for all.

Hey Product Hunt! Eduardo here 👋

Two weeks ago we launched Seagull: real-time translation subtitles for your desktop. The response blew us away.

Today we're back with the feature we've been most excited about: Conversation Mode.

You're sitting across from someone who speaks a different language. Maybe it's a client, a patient, a neighbor, your in-laws. You pull out Google Translate and take turns typing. It's slow, awkward, and kills the flow.

The solution: Just talk. Seagull listens to both of you, identifies who's speaking and in what language, and shows each person the translation in real time. No typing, no turns, no delay.

How it works:

• Pick 2–4 languages

• Hit start

• Have a normal conversation, Seagull handles the rest

Each language gets its own pane in a clean split-screen grid. Speaker diarization figures out who said what. Translations appear as people speak.

Use cases we're seeing:

• 🏥 Healthcare workers talking to patients

• 🏪 Small business owners serving multilingual customers

• 👨‍👩‍👧 Mixed-language families at the dinner table

• 🤝 Business meetings without a hired interpreter

• ✈️ Travelers having real conversations, not pointing at phones

What's under the hood:

• Multi-stream speech recognition with automatic language detection

• Speaker diarization across parallel translation streams

• Temporal speaker remapping to keep IDs consistent

Conversation Mode is available now on Mac, Windows, and Linux. The original subtitle mode is still there too!

Seagull now does both.

We're two developers building this because we think language shouldn't be a barrier to human connection. Would love your feedback.

🐦