Tarannum.ai: Integrating Qari Assessment and AI-Assisted Continuous Learning
Tarannum Technologies
No views · Published October 12, 2026
About the Project
Tarannum.ai connects expert qari assessment with structured practice beyond the classroom through reference audio, live pitch visualisation, recorded attempts and formative technical feedback. Two separate observational evaluations inform its development. In the al-Fatihah study, four qari completed 180 assessments including hidden repeats; 38 unique recordings had complete usable panel ratings. Average-measure absolute-agreement ICC was 0.851, indicating human-panel reliability rather than AI accuracy. Automated scores showed weak alignment with mean qari ratings (Pearson r = 0.080; MAE = 15.89 points), supporting further calibration. In an azan course with 23 attendees, the observation window of 19–24 September 2026 contained 448 eligible recordings. After class, 21/23 attendees (91.3%) continued recording, producing 346 recordings (77.2%). Among 20 participants with at least six scored recordings, mean scores for the first three versus last three attempts rose from 54.66 to 64.49; 18/20 improved. These descriptive technical-score changes do not establish causally improved competence or long-term retention. The innovation integrates expert guidance, repeated practice and recorded evidence for review. Potential adoption includes learner subscriptions and institution-sponsored cohorts, with further evaluation of costs, independent validation and expert-rated outcomes.