Yukon Chang

Audio classification and categorization based on wavelets and support vector Machine

In this paper, an improved audio classification and categorization technique is presented. This technique makes use of wavelets and support vector machines (SVMs) to accurately classify and categorize audio data. When a query audio is given, wavelets are first applied to extract acoustical features such as subband power and pitch information. Then, the proposed method uses a bottom-up SVM over these acoustical features and additional parameters, such as frequency cepstral coefficients, to accomplish audio classification and categorization. A public audio database (Muscle Fish), which consists of 410 sounds in 16 classes, is used to evaluate the performances of the proposed method against other similar schemes. Experimental results show that the classification errors are reduced from 16 (8.1%) to six (3.0%), and the categorization accuracy of a given audio sound can achieve 100% in the Top 2 matches.

Fast, prime factor, discrete Fourier transform algorithms over GF(2m) for 8≤m≤10

In this paper it is shown that Winograds algorithm for computing convolutions and a fast, prime factor, discrete Fourier transform (DFT) algorithm can be modified to compute Fourier-like transforms of long sequences of 2^m-1 points over GF(2^m), for 8=

Robust voice activity detection using perceptual wavelet-packet transform and Teager energy operator

In this letter, a robust voice activity detection (VAD) algorithm is presented. This proposed VAD algorithm makes use of the perceptual wavelet-packet transform and the Teager energy operator to compute a robust parameter called voice activity shape for VAD. The main advantage of this algorithm is that the preset threshold values or a priori knowledge of the SNR usually needed in conventional VAD methods can be completely avoided. Various experimental results show that the proposed VAD algorithm is capable of outperforming the VAD of Adaptive Multi Rate (AMR) speech codec in both additive noisy and real noisy environments.

Investigation of the Relationship between Whole-Wafer Strength and Control of Its Edge Engineering

Silicon wafer breakage has become a major concern for all semiconductor fabrication lines because it is brittle, and thus high stresses are easily induced in its manufacture. The production cost of devices significantly increases even for a breakage loss of a few percent if wafers are broken near completion. Even wafer breakage near the beginning of the process is significant. In this investigation, we develop a brand new approach to reducing breakage by using a charge-coupled device (CCD) to capture the cross-section image of the wafer at its edge; the data measured at the edge can be used to determine overall wafer strength. Analysis of the image of the wafer edge is used to characterize silicon strength, and a simple drop test is conducted to elucidate wafer failure, improving our understanding of the accumulation of stress in the wafer bulk before failure. We also describe many of the improvements that have resulted in the virtual elimination of wafer breakage due to unidentified causes. Our analysis gives the optimal front size (B1), edge widths (A1,A2), and bevel angle (θ) for the edge profiles of wafers to prevent wafer breakage. Briefly, when a suitable material and suitable process control approaches are utilized, silicon wafer breakage can be prevented. This is the first investigation providing evidence that whole-wafer strength is an important issue. We present a physical model to explain why wafer fracture has become an increasingly serious problem as the diameter of wafers has increased. The control of wafer edge geometry has been demonstrated to be an effective means of protecting wafers with large diameters against breakage. This model reveals that the breakage rate of wafers can be reduced by controlling the uniformity of the differences between the front size and the rear edge widths during the wafer manufacturing process.

Coding of arbitrarily shaped objects using pace code

Pace code, a new technique for the shape coding in object-based image technology is presented in this article. It is a lossless, contour-based scheme in which contour pels on the shape contour curve are encoded directly. Similar to the well-known chain code, our method encodes the spatial relation between pels on the neighboring shape contour curve, but it uses a concept called pace which records the relationship differently. Experiment result shows that the proposed algorithm has the best coding efficiency among most contour-based schemes and requires low coding computation.

Identifying Engineering Students' English Sentence Reading Comprehension Errors: Applying a Data Mining Technique.

The purpose of this study is to propose a diagnostic approach to identify engineering students’ English reading comprehension errors. Student data were collected during the process of reading texts of English for science and technology on a web-based cumulative sentence analysis system. For the analysis, the association-rule, data mining technique was applied to mine students’ reading errors. Specific association rules of reading errors were identified and distinctive patterns of error production have been recognized among different groups of students. This article addresses the issue of English reading difficulty frequently found among engineering students and discovered possible tendencies of student reading errors. By using the techniques in this article to identify students’ reading problems, instructors will be able to construct learning materials for adaptive learning, and thus reduce the cost of teaching and facilitate students’ learning outcome. Pedagogical implications were provided based on the results of the study.

Seamless channel transition for broadcatch scheme

BroadCatch belongs to a group of periodic broadcasting schemes that deliver popular video to a large number of users using multicasting. Some of the schemes have been shown to support seamless channel transition, a process that increases or decreases the number of channels allocated to a video without causing service interruption. By reallocating channels among movies depending on their popularity, the performance of a Video-on-Demand service can be improved. In this paper, we propose a seamless channel transition scheme based on BroadCatch and demonstrate its effectiveness as a performance tuning tool with simulation.

Enhancing Engineering Students' Reading Comprehension of English for Science and Technology With the Support of an Online Cumulative Sentence Analysis System

For engineering students, reading in English is the core competence to absorb professional knowledge in academic settings and their future career, because many authentic textbooks and information about advanced technology have been published in English. The present study sets out to improve English reading comprehension among tertiary-level engineering students. An online reading strategy instruction based on cumulative sentence analysis (CSA) was constructed to enhance the students’ reading comprehension of English technology texts. The comparison between the pre-test and post-test showed that the participants achieved a higher level of reading comprehension performance following the instruction. The findings clearly demonstrated that online CSA strategy instruction is an efficient and feasible approach to helping engineering students cope with their problems of reading English texts. Pedagogical implications are briefly discussed based on the findings of this study.

Elucidating the Effects of Current Stress History on Reliability Characteristics by Dynamic Analysis

This work used a new dynamic approach in which wafer-level chip-scale packages (WLCSP) were first aged by applying various current levels and then detailed measurements were taken to study the failure mechanism of these packages. The aging stage involved various accelerated current stressing conditions at a fixed ambient temperature of 125 °C. Unlike previous studies, all of which ran the aging process to the point where failure occurred, our experiment halted the aging stage after a sufficient amount of time to allow us to carefully investigate the failure mechanism in detail. The experimental results show that different levels of stressing currents have different effects on the reliability characteristics of the tested WLCSP packages. This work shows that the dynamic analysis approach is effective in studying the mechanism of WLCSP failure and related reliability characteristics.

Fast classification scheme for VQ

In this paper, a new classification scheme using concentric circles is derived for Vector Quantization (VQ) coder. Two DCT coefficients of each image block are used to compute the radius (r) and angle (θ) in the polar coordinate. A classification scheme consisting of a total of 65 classes is implemented according to the values of r and θ, and is called 65-classes Concentric Circle Classified VQ (65-CCCVQ). Simulation results show that using 65-CCCVQ, the compression speed is much faster than the original VQ algorithm while maintaining similar image quality.