Quality evals
Validate translation quality across models and check how well voice recognition hears you.
Translation quality (chrF + LLM judge)
Runs the built-in German/Ukrainian and English/Ukrainian set through selected models. chrF compares each output with a reference; an optional LLM judge rates adequacy from 1 to 5.
Voice recognition validation (WER)
Read the phrase aloud. Loqui compares what was heard with the reference and calculates the word error rate. Lower is better (0% is perfect).
🇩🇪 Read aloud in German · 1/6
Ich möchte morgen früh nach München fahren.