Bio
Ivana Andjelkovic is a Los Angeles based interdisciplinary software engineer and researcher with a background in data visualization and computer music. Her passion is improving people's music listening experience - be it through working on hi-fi audio products or on music recommendation systems. She holds a Ph.D. in Media Arts and Technology from the University of California, Santa Barbara, M.S. in Computer Science from George Washington University, and a B.S. in Computer Science from the University of Belgrade. She currently works as an Audio Software Engineer at Odeon Inc.
Download my resume
Contact me at iva.andel (at) gmail (dot) com
Recent Work
Audio Processing & Embedded Software
Most recent work at Odeon Inc. includes writing in-house firmware for PIC32 chips to receive and send USB audio data. These chips are now used in a variety of DACs and deliver superior audio quality when compared to previously used chips with licensed firmware. Other projects include research and development of audio processing algorithms for use in hi-fi audio products.
PIC32 chip programmed as an USB audio receiver
Music Recommendation
MoodPlay - Mood based music recommender. The development of this system was motivated by a need to explore music collections by a multitude of words we use to describe moods. Music has a tremendous effect on our emotions, and finding artists whose music feels right at a given moment is not always easy when we face millions of choices.
MoodPlay organizes over 250 moods into a hierarchy and visualizes them in a novel way. The interactive interface built upon the proposed mood model and hierarchical mood filtering make finding relevant artists easier. Furthermore, the system implements a novel, hybrid recommendation algorithm, which adapts to mood changes that naturally occur while listening to music.
The user study on 279 subjects showed the proposed model and algorithm improve satisfaction and recommendation accuracy. Demo video here!
Publications:PDF Moodplay: Interactive Music Recommendation based on Artists’ Mood Similarity. Andjelkovic, I.; Parra, D.; and O'Donovan, J. International Journal of Human-Computer Studies, 2018.
PDF Moodplay: Interactive Mood-based Music Discovery andRecommendation. Andjelkovic, I.; Parra, D.; and O'Donovan, J. In Proceedings of the UMAP Conference, 2016. ACM
Screenshot of the MoodPlay interface
Screenshots of interactive features in MoodPlay: (1) and (2) - the varying recommendation catchment area around user avatar, (3) - trail based interaction
Artist related moods filtered out by their category
MoodPlay system architecture indicating the modules for: (1) dataset construction and (2) recommendation framework.
Visualizations
Notable projects include: (1) dashboard visualizations of data pertaining to the consumption and performance of digital content at Stem Disintermedia and (2) visualizations of large data sets in Research Oriented Social Environment - RoSE. RoSE is an online exploration environment for humanities students and researchers modeled as a dynamic “social network” of authors and works, past or present.
Stem platform: Graph showing breakdown of an artist's earnings for different types of digital content.
Stem platform: Graph showing top performing digital content for an artist.
RoSE platform: Visualization of authors (outer circles), their works (inner circles) and relationships among them (lines).
