iDEA Community: College of Engineering

Content-Based Music Genre Classification Using Sparse Approximation Techniques

Mon, 29 Oct 2012 22:58:59 GMT

Title: Content-Based Music Genre Classification Using Sparse Approximation Techniques

Authors: Aryafar, Kamelia; Adams, Trevor R.; Shokoufandeh, Ali

Abstract: In this study we evaluated the performance of genre classification systems using various feature vectors and learning methods. Using a fixed classifier, i.e., the Gaussian mixture models we were able to create a suboptimal feature vector to characterize the audio signals in a low dimensional feature space. We then utilized this modified feature representation to solve the problem of music genre classification. We evaluated the performance of the recent sparsity-eager support vector machines classifier using the proposed feature vector and compared the results to the classic support vector machines and Gaussian mixture models as the baseline classifiers.

Dominant Color Learning by Subject Extraction

Sat, 29 Oct 2011 22:58:59 GMT

Title: Dominant Color Learning by Subject Extraction

Authors: Aryafar, Kamelia; Attenberg, Josh; Condon, Fiona

Abstract: Advances in the digital media industry have resulted in an exponential growth in available image data sets. This exponential growth has in turn spurred great interest in various methods for acquiring, processing, analyzing, and understanding images in order to produce numerical or symbolic information such as color and texture characteristics. Detecting the dominant color of an object in the image without any prior knowledge about the background model, the object characteristics or the scene geometry is a challenging problem. The two major challenges in assigning a dominant color to the image subject are the isolation of the subject by background subtraction and the extraction of dominant color from the approximated subject region. In this work, we combine an estimated subject mask with the image color histogram to detect the dominant image color.

Automatic Classification of Digital Music by Genre

Sat, 29 Oct 2011 22:58:59 GMT

Title: Automatic Classification of Digital Music by Genre

Authors: Aryafar, Kamelia; Shokoufandeh, Ali

Abstract: Over the past two decades, advances in the digital music industry have resulted in an exponential growth in music data sets. This exponential growth has in turn spurred great interest in music information retrieval (MIR) problems, organizing large music collections, and content-based search methods for digital music libraries. Equally important are the related problems in music classification such as genre classification, music mood analysis, and artist identification. Music genre classification is a well-studied problem in the music information retrieval community and has a wide range of applications. In this project we address the problem of genre classification by representing the MFCC feature vectors in an extended semantic space. We combine this audio representation with machine learning techniques to perform genre classification with the goal of obtaining higher classification accuracy.

Music Genre Classification Using Explicit Semantic Analysis

Fri, 29 Oct 2010 22:58:59 GMT

Title: Music Genre Classification Using Explicit Semantic Analysis

Authors: Aryafar, Kamelia; Shokoufandeh, Ali

Abstract: Music genre classification is the categorization of a piece of music into its corresponding categorical labels created by humans and has been traditionally performed through a manual process. Automatic music genre classification, a fundamental problem in the musical information retrieval community, has been gaining more attention with advances in the development of the digital music industry. Most current genre classification methods tend to be based on the extraction of short-time features in combination with high-level audio features to perform genre classification. However, the representation of short-time features, using time windows, in a semantic space has received little attention. This paper proposes a vector space model of mel-frequency cepstral coefficients (MFCCs) that can, in turn, be used by a supervised learning schema for music genre classification. Inspired by explicit semantic analysis of textual documents using term frequency-inverse document frequency (tf-idf), a semantic space model is proposed to represent music samples. The effectiveness of this representation of audio samples is then demonstrated in music genre classification using various machine learning classification algorithms, including support vector machines (SVMs) and k-nearest neighbor clustering. Our preliminary results suggest that the proposed method is comparable to genre classification methods that use low-level audio features.

HPDLC Films doped with Carbon Nano-onions to improve its electro optic response

Wed, 02 Mar 2011 22:58:59 GMT

Title: HPDLC Films doped with Carbon Nano-onions to improve its electro optic response

Authors: Bellingham, Alyssa; Shriyan, Sameet K.; Fontecchio, Adam K.

Abstract: Holographically-formed Polymer Dispersed Liquid Crystal films (HPDLC) are electro-optical thin films that phase separate upon exposure to an interference pattern to form a Bragg grating composed of alternating layers of liquid crystal droplets and polymer. The Bragg grating allows the film to reflect a preselected wavelength of light. When an electric field is applied to the film , the liquid crystals in the film align in the direction of that field allowing all wavelengths of light to pass through. This switching property makes these films ideal for many applications including displays, biomedical sensors, gas analysis, and Hyperspectral imaging devices. In order to reduce the voltage at which the films switch, materials with high conductivities can be introduced into the polymer regions of the films to slightly increase the overall conductivity of the polymer relative to the liquid crystal. This research focuses on doping the polymer used in the films with carbon nano-onions, which resemble nano-scale buckyballs in structure. They can be functionalized to attach only to the polymer regions during phase separation, which will increase the polymer’s conductivity and lead to a reduction in the voltage needed to switch the samples without hindering the alignment of the liquid crystals. In the mixtures that have been developed so far, there have been aggregates of nano-onions, which reduces the increase in conductivity of the polymer caused by the introduction of nano-onions and decreases transmission through the film. Ideally, the films will retain higher than 80% transmission with the carbon nano-onions, so work is currently being done to reduce the nano-onion aggregates and determine the best ratio of nano-onions to polymer.

Combined Audio and Video Analysis for Guitar Chord Identification

Wed, 14 Apr 2010 22:58:59 GMT

Title: Combined Audio and Video Analysis for Guitar Chord Identification

Authors: Hrybyk, Alex

Abstract: This research presents a multi-modal approach to automatically identifying guitar chords using audio and video of the performer. Chord identification is for stringed instruments adds extra ambiguity as a single chord or melody may be played in different positions on the fretboard. Preserving this information is important, because it signifies the original fingering, and implied “easiest” way to perform the selection. This chord identification system combines analysis of audio to determine the general chord scale (i.e. A major, G minor), and video of the guitarist to determine chord voicing (i.e. open, barred, inversion), to accurately identify the guitar chord.

Description: Student Author: Alex Hrybyk, College of Engineering, ECE; Adviser: Dr. Youngmoo Kim, College of Engineering, ECE

Efficient Acoustic Feature Computation Using FPGAs

Wed, 14 Apr 2010 22:58:59 GMT

Title: Efficient Acoustic Feature Computation Using FPGAs

Authors: Schmidt, Erik M.; Speck, Jacquelin A.

Abstract: Many recent advances in music information retrieval (MIR) have been data-driven. Widespread performance evaluations on common data sets, like the annual MIREX events, have been instrumental in advancing the field. Such endeavors incur large computational costs and could potentially benefit from faster calculation of acoustic features. Traditional cluster-based solutions are expensive and space- and power inefficient. The massively parallel architecture of the field programmable gate array (FPGA) makes it possible to design lower-cost, applicationspecific chips rivaling cluster speed for large-scale acoustic feature computation. Such devices also show potential for implementations of MIR systems on embedded devices where hardware acceleration is a necessity. We present a prototype Xilinx System Generator (XSG) library for acoustic feature calculation. We use a genre classification task to compare the performance of simulated hardware features to those computed using standard methods. Finally, we discuss ongoing efforts toward a working hardware design.

Description: Student Author: Erik M. Schmidt, College of Engineering, ECE; Student Author: Jacquelin A. Speck, College of Engineering, ECE; Adviser: Dr. Youngmoo Kim, College of Engineering, ECE

Musical Humanoids

Wed, 14 Apr 2010 22:58:59 GMT

Title: Musical Humanoids

Authors: Grungerg, David; Batula, Alyssa

Abstract: Humanoids have become increasingly capable in recent years. Enabling these robots to mimic human musical activities is an ongoing area of research; however, most developments in this field have employed pre-programmed motions, and robots remain incapable of responding to changes in music. We have developed algorithms that allow a small humanoid robot, RoboNova, to dance to music and play notes on a keyboard. This robot serves to prototype our algorithms before applying them to Hubo, a more advanced life-sized humanoid. We hope to make Hubo capable of musical interaction, thereby providing a platform to study robot motor control and human creative expression.

Description: Student Author: David Grunberg, College of Engineering, ECE; Student Author: Alyssa Batula, College of Engineering, ECE; Adviser: Dr. Youngmoo Kim, College of Engineering, ECE; Adviser: Paul Y. Oh, College of Engineering, ECE

DrexelCast: Orchestral Performance Companion

Wed, 14 Apr 2010 22:58:59 GMT

Title: DrexelCast: Orchestral Performance Companion

Authors: Hrybyk, Alex; Grungerg, David; Prockup, Matthew

Abstract: Many people enjoy live orchestral performances, but those without musical training may find it hard to relate to the music. We have developed a system that helps users by guiding them through the performance using a handheld application in realtime. Using chroma features and dynamic time warping, we attempt to align the live performance audio with that of a previously annotated reference recording. The aligned position is transmitted to users’ handheld devices and pre-annotated information about the piece is displayed synchronously.

Description: Student Author: Alex Hrybyk, College of Engineering, ECE; Student Author: David Grunberg, College of Engineering, ECE; Student Author: Matthew Prockup, College of Engineering, ECE; Adviser: Dr. Youngmoo Kim, College of Engineering, ECE

The Low to Intermediate Temperature Oxidation of n Propylcyclohexanein a Pressurized Flow Reactor

Wed, 14 Sep 2011 14:39:47 GMT

Title: The Low to Intermediate Temperature Oxidation of n Propylcyclohexanein a Pressurized Flow Reactor

Authors: Corrubia, Julius; Farid, Farinaz; Cernansky, Nicholas; Miller, David

Abstract: Currently computational capabilities for next generation, air-breathing propulsion systems are underutilized in terms of combustion. This lack thereof represents an area of immense research that has ignited a profound interest within the combustion community. However, major hurdles exist that obstruct the community’s pathway to this goal. The important problems that need to be addressed can be grouped into two categories of project goals. First, the combustion properties of practical fuels and their associated surrogate components and mixtures used in air-breathing combustion systems must be understood and quantified. Second, the development of detailed reaction models and strategies for model reduction for use in large-scale simulations must be addressed. These project goals present a daunting task because of the large number of chemical components and classes contained in practical jet fuels derived from petroleum or alternative resources, such as natural gas and coal. It is well accepted that the solution to this problem is to develop surrogates for real jet fuels that contain a reduced amount of chemical components and classes. These surrogates are developed to match the physical properties and chemical kinetics of the practical jet fuels such that the combustion phenomena of the surrogates mimic that of the real jet fuel. Currently the combustion properties of practical jet fuels remain poorly understood and surrogate development is an ongoing process. The desired outcome of this effort is the improved qualitative understanding and quantitative predictability of the combustion properties of practical jet fuels and their surrogates, and the development of reliable kinetic models that may be used in practical combustion applications for design purposes. The JP-8 jet fuel cycloalkane surrogate component n-propylcyclohexane was oxidized in the Drexel Pressurized Flow Reactor (PFR) to gain further insight into its associated combustion kinetics.

Description: This poster was presented at Drexel University Research Day 2011.