Mink Backend v2.0.0 is out! 🎉

We’re excited to announce the release of the Mink Backend v2.0.0 — and while you might not notice huge differences on the surface, behind the scenes we’ve given the whole thing a serious refresh.

What’s new

• 🗣️ Speech recognition: You can now use audio files as the sources of a corpus. Mink will then use Whisper to transcribe them to plain text before running the usual analysis.
• 🧩 Under the hood overhaul: Mink’s backend now uses FastAPI instead of Flask. This brings cleaner code, easier maintanance, and a more modern foundation.
• 📚 Improved API documentation: The new backend automatically generates detailed API docs, which will help developers who integrate directly with Mink.
• 🧠 Better internals: We’ve improved cache management, exception handling, timestamp handling, and introduced automated tests.
• 🧹 Clean-ups and consistency: Some parameter names and routes have been updated or deprecated to make the API more consistent.

For more detailed info you can check the release notes here:
👉 github.com/spraakbanken/mink-backend/releases/tag/v2.0.0

A new version of the web interface was also released, with mostly small adaptations to the Backend changes:
👉 Mink Frontend 1.13.0

What this means for you

If you usually use Mink through its web interface, the only change you should notice is the added ability to analyze speech data. Think of the other changes as a tune-up rather than a remodel. You can expect the same experience you’re used to, but with a sturdier foundation that will make future updates smoother and more reliable.

For those who work directly with the Mink API, there are a few name and route changes to note. Please check the release notes for details.

Thanks for using Mink, and happy processing!