Evalyz: AI-assisted assessment tool of interpreter proficiency

Background

In critical situations such as trials, medical emergencies, and meetings with public authorities, high-quality interpreting can be essential for protecting people’s rights and ensuring access to important services. However, current approaches to assessing interpreter proficiency largely rely on manual expert evaluation, making the process time-consuming, subjective, and difficult to scale. This often leads to inadequate or absent assessment, and as a result, less qualified interpreters are often assigned to tasks they are not equipped to perform.

Development of Evalyz

Evalyz aims to solve this problem by investigating how AI and language technology can support a more systematic and data-driven approach to interpreter assessment, thereby addressing the problem of interpreting quality in high-stakes settings. The project, led by postdoctoral researcher Yao Zhang, aims to develop an AI-assisted tool to analyze and interpret performance and provide structured feedback for training.

The project builds on research in interpreting studies and speech analysis. By combining insights from linguistic research with advances in AI, the tool explores how key aspects of interpreting quality – including fluency, accuracy, completeness of information, and language quality – can be measured in an objective, consistent, and transparent way.

The project has mainly been funded by Spinouts Denmark and has received additional Proof-of-Concept funding from UCPH Lighthouse. 

Impact and futher use of Evalyz 

The project aim was to supplement human judgment by providing interpreters, educators, and language service providers with an additional tool to support more efficient, data-informed training, assessment, and continuous improvement across interpreter training, recruitment, and professional services. The long-term vision is to develop a scalable quality assurance infrastructure for the interpreting sector, supporting more reliable assessment practices across interpreter education, recruitment, and professional services.

The project is currently developing an MVP (Minimal Viable Product) which means a version of the platform with just enough features to be usable by early customers who can then provide feedback for further development. In this process, the project is collaborating with stakeholders in the interpreting sector to explore future applications.

Project partners

The project has been led by post.doc. Yao Zhang, who has been employed at CIP for the period of 1 April 2025 - 30 June 2026.