The Information and Language Processing Research Lab (ILPRL), Department of Computer Science & Engineering, Kathmandu University(KU), Central Campus, Dhulikhel, released the Google Trilingual Machine Translation (TMT) System on December 26, 2025, for community-level testing and feedback. The system available through the website, https://tmt.ilprl.ku.edu.np, currently has four models which can translate in the following directions ( Nepali-Tamang, Tamang-Nepali, English-Nepali, Nepali-English, English-Tamang, Tamang-English). After a month of gathering feedback and possible corrections/suggestions from the users on the output of the models, the best-performing model will be selected for final deployment. One of the thirty selected recipients of the Google Academic Research Award 2024 worldwide under the Society-Centered Artificial Intelligence category, the project aims to enable access to the information, enhance communication, and preserve minority languages like Tamang.

“We have now almost completed the technical aspects of the Project and have entered the third phase, which is essentially community engagement, under which we will reach out to the community and help them familiarize the system by means of training and workshops after the final deployment of the system. We will also simultaneously release the Corpus Annotation Tool, which will allow Tamangs globally to contribute with translations, thereby ensuring community-level participation in the overall enhancement of the system”, said Prof. Dr. Bal Krishna Bal, who is the Principal Investigator of the Project and the Lead Researcher of the ILPRL Lab, KU. The Tamang Rastriya Pustakalaya, led by Mr. Amrit Yonjan-Tamang and his Tamang Translation Team, plus Prof. Dr. Balaram Prasain, who is also the co-Principal Investigator of the Project, have provided the linguistic and translation support to the Project.

Currently, the system has been trained just on 23 thousand parallel sentences from five different domains, viz., Agriculture, Health, Education, Culture & Society, and General Communications, but in the final deployment, the aim is to go up to 100 thousand parallel sentences. “We will soon approach Google to officially enlist the Tamang language in the Google Translate application. This will be a major step towards internationalizing the language,” said Prof. Bal.

“This TMT system will be very helpful in terms of supporting e-governance and government service delivery both to the Government and the citizens”, remarked Prof. Prasain and Senior Tamang Linguist Mr. Tamang.

The event was chaired by the Chief Guest, Prof. Dr. Manish Pokharel, Dean, School of Engineering.

tmt.ilprl.ku.edu.np