Validate first
Before anything is analysed, the recording is checked for usable audio: is there any signal, is it long enough, is the room drowning it out. Empty or unusable recordings are returned for a retry and are not counted against your allowance.
A score per sound, not per attempt
A single number for a whole word tells you something went wrong without telling you where. The breakdown attributes the result to specific sounds in specific positions, which is the part you can actually act on.
Keep feedback readable
Every score is paired with a written cue in plain language, so the feedback loop never depends on hearing a playback. If it matters, it is on the screen in text.
Practise with intent
Patterns across your history surface the sounds and words worth revisiting, so the next session has a target before you sit down instead of being chosen while you are already tired.
What the score is not
The score is a practice signal, not a clinical measurement. It is affected by microphone quality, background noise and accent. It does not diagnose anything and should not be treated as an assessment.
What happens to the recording
Audio is uploaded, validated, analysed, and turned into a score. The derived scores stay as your progress history. Raw audio is retained only for the window described on the privacy page, and you can request deletion at any time.