Home | News | 04.10.2024

On 9 October 2024 (Wednesday) in 14:00 the hour, in the premises of the Rectorate of Ss. Cyril and Methodius University in Skopje, The launch of 'Buki' will be held – an open, AI-based digital model for speech recognition and transcription in the Macedonian language.

The model was created by Dejan Porjazovski, an expert in artificial intelligence technologies for speech recognition from Aalto University in Finland, Prof. Dr Nikola Stikov from the Technical University of Montreal, Canada and Prof. Dr Ordan Chukaliev, as part of the activities of the Centre for Advanced Interdisciplinary Research (CeNIIS) at UKIM. The model enables the transfer of spoken speech into written text, which includes punctuation, lower- and upper-case letters.

Currently, the literary language model is operational, but a second model is also being developed that would recognise the Macedonian dialects. At the launch, a comparison will be made between the new Buki-Whisper (an improved version of Whisper) and the largest open-source models currently available – Whisper (OpenAI) and Meta (Facebook).

The training of 'Buki' used original data provided by Prof. Dr. Ilina Jakimova, co-head of the Institute of Ethnology and Anthropology at the Faculty of Natural Sciences and Mathematics, UGMK, and from: the Digital Archive for Ethnological and Anthropological Resources (DAEAR) of the Institute for Ethnology and Anthropology, Faculty of Natural Sciences and Mathematics at UCM; the audio version of the international journal 'EthnoAnthroZoom' of the same institute; the audio podcast 'Ordinary People' by Ilina Jakimova; the educational videos from the 'Science for Children' series, the KANTAROT Foundation, and the Macedonian version of Mozilla Common Voice (version 18.0).

After the launch, the digital model will be open for testing. Anyone who wishes to contribute will be able to donate their voice – by reading a text in Macedonian and uploading it to the Mozilla platform, for which more detailed information will be shared at the launch. To that end, UКИМ will launch a campaign "„Donate a voice“.

The new digital platform can benefit everyone who works with text transcription (interviews, lectures, field notes, reminders), people with disabilities, the transcription and annotation of Macedonian cultural heritage, archives, recordings at MRTV, etc.

The name "Buki" derives from the name of the second letter of the Glagolitic alphabet Ⰱ ("buki") – meaning letter or letter. In English, the same will be transcribed as "Bookie.".

The link to test the digital model will be provided during the launch.

Ss. Cyril and Methodius University in Skopje