I built an app that translates WhatsApp voice notes and lets you reply by voice in their language

My family sends voice notes. A lot of them. Two minutes of Punjabi from my mom, a quick one in Hindi from an uncle, and I understand maybe half of it.

WhatsApp can now transcribe voice messages on some phones, but only into text in the same language. That doesn't help when the problem is the language itself. And typing my reply into Google Translate, then pasting back text they won't read anyway (they prefer voice), never felt right.

So I built ReplyLingo, a small app for iPhone and Android that does three things:

1. Understand. Share a voice note from WhatsApp or Telegram into the app. It transcribes it and translates it into your language.
2. Listen. It can read the translation out loud.
3. Reply by voice. Record your answer in your own language, and the app turns it into a voice message in their language that you send straight back.

That third part is the one people react to. Parents hear a voice reply in their own language instead of a block of English text.

The stack

Nothing exotic, but a few choices might help other solo builders:

• Expo / React Native for one codebase on iOS and Android. The share-sheet integration (receiving an audio file from WhatsApp) took the most fiddling on both platforms.
• Whisper for speech-to-text. It handles messy, real-life voice notes (background noise, mixed languages) better than anything else I tried.
• An OpenAI model for translation. It keeps tone and casual phrasing much better than word-by-word translation, which matters for family chat.
• Google Cloud Text-to-Speech for the voice reply, with ElevenLabs as a fallback when a language or voice isn't supported.
• Firebase for auth, RevenueCat for subscriptions on both stores.

Things I learned

1. "Translate" means different things to different users. Some only want text in their language. Others mostly want to reply.

2. Store listings matter more than I expected. People looking for an app like this search the App Store and Play Store, not Google, so I localized both store pages (keywords, descriptions, screenshots) into many languages.

3. TTS fails in surprising ways. Some language and voice combinations just don't exist in one provider. A fallback provider saved a lot of broken replies.

4. Free trials are not automatically good. After I added a 3-day trial, people who used to buy straight away started a trial instead. I'm still testing the right setup.

Try it

ReplyLingo is free to start on the App Store and Google Play. More at replylingo.com.

If you've built anything with voice, translation or share-sheet extensions in React Native, I'd love to compare notes in the comments, and I'm happy to answer questions about the stack.