Or, my Google Scholar page

Papers (2018- )

Papers (2014-2018)

4+1 pages, as the title

4+1 pages. Why don’t we retrieve drum sample sounds with your voice query?


6+3-page conference paper. “My convnet features are better than low-level audio features + MFCC + rhythmic features + a lot of them!” for 6 tasks such as genre classification and emotion prediction.

A beginner’s guide for deep learning x MIR

2+1 workshop article; easier prototyping your neural network for audio signal.

10-page, a journal paper submission. How much and why music tagging datasets are noisy? What’s the effect?

5-page conference paper. Conv+RNN (or CRNN) for music tagging with less number of parameters. I compare three structures while changing the number of parameters.

3-page extended abstract: If music playlists are automatically generated, so should descriptions for them, because otherwise listeners would be puzzled.

7-page paper: It has 88 spectrogram images as I had to explain how it’s like to listen 4 songs * 64 features * 5 layers = 1280 music excerpts from a ConvNet features.

4-page workshop paper: Why don’t we use internal transitions as references of between-tracks transitions – and let RNN model them? Although it’s lack of proper evaluations, yet.

7-page paper: Deeper ConvNet for music tagging because it outperforms. One of the reviews said it’s ‘result-driven’ and actually I agree with it. Still it has some nice conclusions – e.g. melspectrograms > others.

8-page single-column paper: Text-based RNN to generate drum tracks and jazz chord progressions

2-page extended abstract: If we can visualise the learned features, we can listen to them!

3-page workshop paper: I was curious about things like serendipity and similarity of music playlists. But the work is quite simple, please take it with a grain of salt. Well, Lille was a nice city!

Papers on audio/music (2010-2014)

  • Multichannel-to-Wave Field Synthesis Upmixing Technique Based on Sound Source Separation 
    Keunwoo Choi, Tae-Jin Park, Joengil Seo, and Kyeongok Kang, 2013, 52nd AES Conference, Audio Engineering Society pdf
  • Harmonic and Percussive Separation Based on NMF and Tonality Mask 
    Keunwoo Choi, Sang Bae Chon, and Kyeong-Ok Kang, 2012, Journal of ETRI Letter pdf link
  • A System for Korean Traditional Music Notation System : A Definition of XML File Format and An Editor Application
    Keunwoo Choi, Tae-Jin Park, Yongju Lee, and Kyeongok Kang, 2013, 135th AES Convention, Audio Engineering Society pdf
  • A NMF-Based Mono-To-Stereo Blind Upmix
    Keunwoo Choi, 2011, Master’s Thesis pdf