Run Umuseke on a real sample.
Pick a scenario, choose a target language, and the console calls our API for the result. These are curated reference outputs rather than live inference — the response says so, and so do we.
Want to run this on your own data?
Send us a sample — dialect, field recordings, scanned forms. We will run it and share what the model got wrong as well as what it got right.