I have worked in industry and academia. While working with the digital musical instrument division at the LG High Media Lab, I have been involved in implementation of auto accompaniment module, sound programming, debugging and manual writing for the GS3000 digital musical workstation.  I have also been involved with voice codecs, DTMFs, VAD and analysis of communication layers for GSM mobile communication technology at LGCIT. See publications section for related papers. Below are examples of recent and on-going research projects.

Electro-Acoustic Music Mine (EAMM) 2011 ~
Research and development of crowd-source-based electro-acoustic music archives. Project is divided into developing servers in 3 phases (1) filtered crowd-sourcing, (2) archiving, and (3) MIR-based music analysis and visualization.

Systematic and Quantitative Electro-Acoustic Music Analysis (SQEMA ) 2010 ~
Research in systematic and quantitative methodologies for electro-acoustic music analysis. Quantitative analysis is achieved via the EASY Toolbox.

Electro-Acoustic Music Software Analysis (EASY) Toolbox 2008 ~
Computational analysis of electro-acoustic music. Research in electro-acoustic music analysis and development of software using MIR (Music Information Retrieval) techniques including salient feature extraction, segmentation, and pattern recognition algorithms. The program is currently being developed in MATLAB and will be ported to platforms for efficient signal processing and 3D visualization and multi-touch interaction with the GUI system.

FMS (Frequency Modulation Synthesis) 2007 ~
Research and development in synthesis method using a re-synthesis by analysis approach for modulating sound objects’ timbral dimensions based on salient feature extraction techniques. Resulted in a number of papers and numerous invited talks in the US as well as Korea, China, and Russia. Project is currently implemented in MATLAB and will be ported to platforms that exploit multi-touch control of timbral spaces for resynthesis.

Matlab/Simulink

• Simple(st) Karplus Strong Model Matlab code for simple plucked string using noise burst, IIR filter.  Get code here.

• Extended Karplus String Model (Matlab/Simulink) Implementation of the "Plucked String Model". Get code here

• Pitch Detection(Matlab) Basically uses autocorrelaion, cubic spline interpolation, time averaging of peaks and fft  for pitch identification. Currently refining.

• Additive Synthesis Uses amplitude envelope and frequency envelope information for synthesis of time varying harmonics. You can hear a little demo of a trumpet using a data set from 'Elements of Compuer Music', Grey 1975 pg. 218. Get code here.

• Peak Picking Searches for peaks in an FFT spectrum. Returns the locations and magnitudes of the peaks both in linear and logarithmic magintue spectrum. (Not necessarily harmonic relationships, soon to be updated after some major revamping). Get code here.

• Spectrogram and Peak Picking of Spectrogram Does a spectrogram using the STFT and returns an array containing frame by frame spectrum information. The spectrogram's output is used to do a peak search on a frame by frame basis extracting salient peaks that are in the signal.  "Saliency" is controllable via input parameter.  Get code here.

• Phase Vocoder, control of windows size, hop sizes and window type.  Get code here.

 
• Traveling Circus, using Durbin-Willshaw algorithm to find contour of the travelling circus problem. Get code here.

 
• Finding Clusters, using the EM method. Get code here.

 
• XOR - Backpropagation neural network, using EBF(Elliptical Basis Function) and  RBF(Radal Basis Function). Get code here.

• Internet Telephone Application. Using 13 kbps voice codec. Integrating front end, codec and network modules... Please wait for final version.

• "Salient Feature Extraction of Musical Signals"  was written in Java which includes Swing and Java Sound. This software is a package that includes signal processing modules that extract salient features from musical signals and display them on a GUI interface. The program will soon be posted here.

• Mighty Mouse. Use of alternate sensors to investigate usefulness of the standard mouse.  FSRs (force sensing reseistors) and photo-resistors with a Basic Stamp II chip fitted into a standard mouse.  The interface (GUI and serial communication on PC side) is written in Java. Get code and report here.
• LoMaHa (LookMaNoHands). The power of piezo and misc. sensors "at the tip of your hands".

BeOS Related

• Digital wah-wah effect processor. Control over LFO, Q, sweep speed, input/output gain. Please wait for upgrade to current OS.

• Algorithmic piece (Circle Hana), please wait for upgrade to current OS.

• Voice codec. 13kbps @ 8 kHz. Written in ANSI C. Executable DOS file, input is binary floating point file. LPC using Schur recursion with preemphasis and DC offset compensation, get code here.