dc.description.abstract |
In a world full of sounds, music is used as a connection for people around the
world. This proposed project titled "Music-Source-Separation" is a deep neural network
reference implementation for creating opportunities for researchers, audio engineers, and
artists. The completion of the project will result in providing a genuine and clear concept
of the sound and instruments. This project deposits music and permits users to separate
pop music into four stems: vocals, drums, bass, and the remaining other instruments. The
user can observe the sound from the sequence that is inserted from the system. A lengthy
history of music separation has a scientific interest because of being thought of as an
immensely difficult problem. For example, deep learning-based systems have been giving
very meaningful separations which lead to increase interest commercially. MusicSource-Separation giving a reference implementation where a deep neural network is
basically established. The project itself provides two main purposes. Firstly, accelerating
all academic research in this field. Secondly, Improving the Bengali music community. The
The Bengali music community has a bright history throughout the birth of the nation.
Research for Bengali music has been very poor throughout the years. The artists also
suffer from various problems during their careers and research has not been done for their
work in their life. There is not a huge amount of collections from the past and also not
enough resources created for future endeavors. Our work will try to affect the Bengali
music community by creating a healthy and educative environment for all ages of people. |
en_US |