How it works

You speak, janqua listens, Aria coaches

Learners choose passages in their target language from a large library graded by complexity, and read them aloud. janqua analyses and scores their speech against native speakers for the target language. Then Aria and her team of AI coaches provide very detailed feedback and design drills around any weaknesses.

1. Assessment

Learners read graded passages matched to their level, from beginner to advanced. janqua analyses this input against native language properties including phoneme accuracy, speech flow, rhythm and stress and, for tonal languages like Cantonese and Mandarin, tone.

Learners get an overall score as well as a score for each of these properties. The overall score is broadly aligned with the six levels (A1 to C2) of the widely recognised Common European Framework of Reference for Languages (CEFR).

2. Analysis

janqua's analysis goes much deeper than apps that simply ask the learner to "try again", with no insight into where they missed the mark. Every assessment produces a detailed report with a full linguistic breakdown: sound-by-sound and word-by-word analysis and scoring, charts comparing the learner's pitch and rhythm with a native speaker's, and a replay of the passage in the learner's own voice, adjusted to simulate the target pronunciation.

3. Improve

The analysis sets a baseline for improvement. This is where learners meet Aria, our lead AI coach, and her team of specialist AI language coaches. Aria assigns coaches based on each learner's analysis results.

The coaches give detailed feedback on how to improve, including the changes to jaw, tongue and lip position that are critical to clear speech. They set targeted drills with immediate scoring and feedback. Once learners hit their targets, they can take a new assessment to review their overall progress.