Methodology & Safety Statement

How Dia AI Evaluates Speaking in a Safe Environment

Our pedagogical model is designed around active output, transparent metrics, and strict user safety covenants.

Pedagogical Philosophy

Mentora is structured around common classroom English competencies. We believe that learning language is an active, conversational behavior rather than a passive list of multiple-choice questions.

Our lessons are inspired by school grammar needs and suitable for classroom-style English practice. By combining structured writing, vocabulary exercises, and spoken dialogues, we support CEFR-style skill development across grammar tenses, vocabulary range, and pronunciation.

Speech Evaluation Mechanics

Dia processes spoken answers locally. When a student talks during a speaking mission, Mentora converts speech into text and compares it to expected grammar rules and phonetic pronunciations.

We evaluate response accuracy, syntax, and delivery confidence, immediately highlighting word corrections. This allows learners to iterate on speech patterns instantly in a non-judgmental environment.

Privacy & Zero-Audio-Storage Guarantee

We prioritize voice safety. Mentora uses browser-native speech recognition technologies to perform evaluation where possible.

We enforce a strict zero-audio-storage policy on our servers: we do not save, record, or retain any raw audio or voice recordings. Only compact scoring metadata and transcript strings are sent to compute performance percentages, preserving data minimization.

In addition, parent profiles feature filtering for sensitive assessment and speaking details, ensuring that detailed performance lists are restricted to appropriate viewers.

Start specialized classroom English practice today

Gain immediate conversational confidence in safe, structured interactive modules.