The SAGE project (Spoken Assessments Guided by Enhanced technologies) is a cutting-edge AI-driven research initiative under the imec.icon program, jointly funded by imec and the Agentschap Innoveren & Ondernemen (VLAIO). It aims to overcome a major gap in current digital assessment practices: reliable, scalable, and fair evaluation of spoken language performance. While digital tools have transformed written assessments, spoken assessments have largely remained manual, slow, inconsistent, and potentially biased, especially in multilingual and multicultural settings.

At the heart of SAGE is the development of trustworthy AI technologies that help human assessors grade spoken responses more efficiently and transparently. These technologies combine advanced speech analysis, transcription, language evaluation, and explainable machine learning methods to deliver consistent scoring, meaningful feedback, and deeper insight into both form and content of spoken responses. The project tackles core technical challenges, such as handling diverse accents and speech patterns, producing accurate transcriptions, and generating automated yet explainable evaluation rubrics that can highlight fluency, content relevance, and response quality.

SAGE’s research outputs are being validated in three real-world use cases:

• High-stakes oral exams (e.g., biomedical dialogue scenarios in English),
• Summative spoken tests in STEM domains (monologue delivery in Dutch),
• Formative language practice apps (interactive spoken practice in French).

These demonstrations are designed to ensure that the technologies are not only academically sound but also practical and scalable for educators, language trainers, and professional assessors. Ultimately, SAGE seeks to make spoken assessments fairer, more efficient, and more transparent, reducing bias and expanding access to high-quality evaluation tools while unlocking new opportunities in educational technology markets.

The project started on October 1, 2025 and will run until September 30, 2027, involving a consortium of industry and research partners including BLCC, Linguineo, Sensotec, Televic Education, EdTech Station, and research groups from UGent and KU Leuven.

‍

More info: www.imec-int.com/en/research-portfolio/sage

                  www.imec.be/nl/vlaamse-innovatiemotor/samenwerking/imec-icon/projectportfolio/sage

‍