Introducing MARG

Music and Audio Research Group or MARG at Seoul National University is a highly interdisciplinary research group that tries to find solutions to a number of intriguing problems in audio/music that still remain unsolved, such as:

“How to search/discover songs of my taste from 10’s of millions of tracks online?”
“I’d like to extract just a bass guitar part from Thinking Out Loud for my guitar practice.”
“I got chills every time I hear a guitar solo from When the Levee Breaks. Why?”
“If I could only listen to my favorite songs with a voice of Freddie Mercury or David Bowie…”

With digital audio signal processing and machine learning as two main tools, we have been tackling these problems: the best part is, as we learn more about human’s auditory perception/cognition, the more interesting problems we discover. 

So, if you love music and if you have ever wondered about the above questions and the like, you’re more than welcome to join our group!

What are people doing at MARG?

View all publications

What are people doing at MARG?

View all publications

In Music App

“How do I explore new music of my taste? I am not interested in what is recommended by my online music service.”
“I don’t remember the title or artist name of the song I listened to the other day. All I remember is the melody line from the chorus part. How can I find it?”

  • Recommender Systems

  • Music Retrieval

When Listening Music

“I want to remove the vocal part only from any recordings for singing practice. Would it be possible?”
“Whenever I listen to ‘Fake Plastic Tree’ by Radiohead, I shed tears. What’s going on inside my brain?”

  • Source Separations

  • Music Perception/Cognition


“I want to play ‘Guitar Hero’ but with my REAL guitar. And I want to choose whatever songs I like from my collection. Can I do it?”
“What if I could play flute without using my mouth? How about with my whole body?”

  • Audio/Music Signal Processing

  • New Interface/Interaction Design


Currently active research projects at MARG

Hear what you Imagine

Cross-modal audio synthesis/transformation using generative adversarial network

Learn more


Auto-regressive Generative Adversarial Network for Singing Synthesis and Evaluation

Learn more


Development of smart sleep mask using acoustic biomarker and machine learning

Learn more

Lyrics Sync

Development of Automatic Lyrics Synchronization Algorithm Using Deep Neural Network

Learn more

Do you want to see more research?

Go to Research

Latest News

Latest News

Check our news

Contact us

Music and Audio Research Group
443-270, 864-1 Iui-dong, Yeongtong-gu,
Suwon-si, Gyeonggi-do, Republic of Korea
TEL +82 031 888 9150