From Voice to Verdict: An AI-Based recitation app for judicial Competitive Exams

Abstract

This project entails the development of a desktop application to facilitate the study of justice administration civil service exams (particularly for judges) using voice-based techniques. The system allows users to “recite the law” aloud (speak memorized legal texts) and leverages Natural Language Processing (NLP) to evaluate the accuracy of the recitation. The application compares the speech-to-text transcription of the user’s voice against the official law text and computes a percentage of correctness, giving immediate feedback on errors and omissions. This undergraduate thesis documents the project motivation (born from an interest in mobile app development and AI in education), the project management and methodology used, the analysis of technologies (speech recognition, development frameworks, text comparison techniques), implementation details (system requirements, UX design, data persistence, unit testing), and a discussion of results. The work concludes that the tool helps improve candidates’ performance by offering an innovative way to practice oral exams, meeting the stated objectives and paving the way for future improvements in accuracy and features.

Aquest projecte consisteix en el desenvolupament d’una aplicació mòbil per facilitar l’estudi d’oposicions de justícia (especialment per a jutges) mitjançant tècniques de veu. El sistema permet als usuaris “cantar la llei” (recitar en veu alta textos legals memoritzats) i utilitza processament del llenguatge natural per avaluar l’exactitud del recitat. L’aplicació compara la transcripció de la veu de l’aspirant amb el text oficial de la llei i calcula un percentatge d’encert, proporcionant feedback immediat sobre errors i omissions. Aquest treball de fi de grau documenta la motivació del projecte (sorgida de l’interès en la informàtica mòbil i la IA aplicada a l’educació), la gestió i metodologia emprades, l’anàlisi de tecnologies (reconeixement de veu, frameworks de desenvolupament, tècniques de comparació de text), els detalls d’implementació (requisits, disseny UX, persistència de dades, proves unitàries) i una discussió de resultats. Es conclou que l’eina ajuda a millorar el rendiment dels opositors en oferir una forma innovadora de practicar els exàmens orals, complint els objectius plantejats i obrint la porta a futures millores en precisió i funcionalitats.

This project entails the development of a desktop application to facilitate the study of justice administration civil service exams (particularly for judges) using voice-based techniques. The system allows users to “recite the law” aloud (speak memorized legal texts) and leverages Natural Language Processing (NLP) to evaluate the accuracy of the recitation. The application compares the speech-to-text transcription of the user’s voice against the official law text and computes a percentage of correctness, giving immediate feedback on errors and omissions. This undergraduate thesis documents the project motivation (born from an interest in mobile app development and AI in education), the project management and methodology used, the analysis of technologies (speech recognition, development frameworks, text comparison techniques), implementation details (system requirements, UX design, data persistence, unit testing), and a discussion of results. The work concludes that the tool helps improve candidates’ performance by offering an innovative way to practice oral exams, meeting the stated objectives and paving the way for future improvements in accuracy and features.