Learning a foreign language is like tuning a musical instrument. The new technology works like a smart tuner: it takes your recording and a native speaker's reference speech, stretches or compresses them in time (like adjusting a string) to align them precisely, and evaluates three key parameters: clarity of sounds, rhythmic precision, and intonation. No transcripts or annotated examples are needed — the system itself breaks down speech into frequency components (spectroscopy) and compares them to a Standard Model — the native speaker's recording.

The result is surprising: the rhythm assessment, based on analyzing temporal deformations (entropy of the speech flow), matches professional experts' opinions even more often than their assessments match each other. The idea traces back to the work of Фурье, who showed how complex oscillations reduce to simple waves — essentially, the notes that make up any melody. This approach is universal — it is equally accurate for English and Japanese, which means it will ease the learning of rare languages where educational materials are acutely lacking.