KachaKache (කච-කචේ) v0.1.0 is now live.

I’ve been using tools like Wispr Flow for quick voice-to-text.
But two things kept bothering me:

• It’s cloud-based
• It’s a monthly subscription

So I thought…
we already have solid open-source voice models — why not just run it locally?

That idea turned into a small weekend project.

What KachaKache does:
• 🎙️ Voice → Text (fully local)
• 🔒 Privacy-first (no data leaves your machine)
• ⚡ No account, no login
• 🧠 Supports 4 voice models (base en works great for me)
• 🧹 Built-in lightweight rule engine for cleanup

I’ve also shared a quick video showing how it works.

Since this is an unsigned macOS app, you’ll need to run this once before opening:
xattr -dr com.apple.quarantine/Applications/KachaKache.app

If that feels sketchy (fair enough 😄),
you can always clone the repo and build it locally.
This is still v0.1.0 — very early.
There will be rough edges.

But if you’re someone who wants:
→ local-first tools
→ privacy over convenience
→ no subscriptions for basic workflows
…this might be useful.

If you try it out, let me know what you think.
Ideas, improvements, weird bugs — all welcome.
PRs are open too

KachaKache

Speak naturally. Get cleaner text.

Speak naturally. Get cleaner text.