A random word is shown with its IPA transcription. Record yourself saying it — the model will show your pronunciation colored by confidence.