Pick a situation and press Start. Lena speaks German; answer by voice or by typing. Your mistakes are corrected under each of your messages and saved to “My mistakes”.
Listen for these
Now retell it in your own words
Generate a short workplace audio, listen without the transcript, then retell it aloud. You get scores, corrections and a model answer.
Hear one sentence, repeat it (or type it as dictation), then check. Works with the last “Listen” text or any German text you paste.
Listen, say it, and see word by word what the app understood. Green = clearly understood, yellow = understood but unclear, orange = heard a similar word, red = heard something else or nothing.
All saved mistakes
Install on this device
AI tutor
Choose which AI service plays Lena, writes the audios, scores your answers and translates. Speech recognition and the German voice are not affected by this.
Speech recognition (optional)
Normally the phone's own speech recognition is used. With a Groq (free tier) or OpenAI key, Whisper can be used instead: set “Speech” at the top to “Whisper”. It works in every browser and lets you replay your own recording in the Pronunciation tab. If your AI service above is already Groq or OpenAI, its key is used and you can leave this empty.
Give it a name to keep it in “Saved setups”, so you can switch between keys and services with one tap.