Audio Source Separation

In many audio applications, separating individual audio sources from a mixture of audio signals is critical, as it can enhance audio quality or enable precise audio editing. For instance, a singer's voice can be isolated from the instruments, or one speaker's voice can be separated from the other speaker's voice in a conversation. Our EECS 351 final project focuses on developing a system that separates distinct sound sources from complex music audio recordings.

 

We propose to develop an audio source separation tool that utilizes publicly available audio datasets. We will implement digital signal processing (DSP) techniques in Python and MATLAB to separate audio components using a variety of methods.