69 teams from all over the world took part in the competition, including teams from the USA, Belarus and Ukraine. The competition was a two-stage event. In the first stage, reports were evaluated. 25% of the best projects were qualified to the second stage, where teams presented their projects. According to the jury, the team of students from Lodz University of Technology took first place, together with students from Belarus and Ukraine. As a reward, the winners' works will be published in the Central European Researchers Journal.
FriML is a tool for generating music, using machine learning algorithms, made available as a web application. The aim of the project was to prepare a complete solution that would allow the end user to generate new melodies with just a few mouse clicks. The solution was based on LSTM recurrent networks, used for tasks requiring predicting sequences of data on the basis of a certain context. This includes text classification, speech and image recognition, sentence generation and music," explains one of the student mentors - Dr Eng. Piotr Duch from the Institute of Applied Computer Science.
FriML has many potential applications. It can help artists, developers and video game designers to soundtrack their creations. This is extremely easy to do, thanks to the clear user interface of the ready-made application, which runs in a browser. The application enables generating melodies belonging to different musical genres, with different keys and instruments.
The students who developed the project belong to the Main student research group operating at the Institute of Applied Computer Science at Lodz University of Technology and the application can be tested at https://friml.netlify.app/. Due to hosting limitations, this version is strictly demonstrative - the generated melodies last a few seconds and can only be saved in MIDI format.