2003 March, Vol 51 Number 3 |
CONTENT
PAPERS
Signal Representation Including Waveform Envelope by Clustered Line-Spectrum Modeling
M. Kazama, K. Yoshida, and M. Tohyama 123
When modeling a narrow-band signal, which includes spectral lines close to the peak, the effect of windowing truncation also appears in the region near the peak spectral line. By extending previous research, the authors propose a method to model the spectrum near peaks. This requires an interpolation to find the peak, peak picking, and a model of the effect of windowing in order to extract the slowly varying envelope information. Several examples of audio signals with noise, called clustered line-spectrum modeling, are used to demonstrate practical applications of the approach.
ENGINEERING REPORTS
A Moving-Horizon Optimal Quantizer for Audio Signals
Graham C. Goodwin, Daniel E. Quevedo, and David McGrath 138
Quantizing, or requantizing a previously digitized signal, produces errors that are perceived as noise. While a time-invariant noise-shaping feedback can move the noise spectrum to a region of less sensitivity, this conventional approach is actually a subset of a more general method. In the proposed extension, quantization decisions use a perceptual model that also analyzes future samples. A look ahead of only three samples, called a moving horizon, is sufficient to produce even lower perceived noise without adding significant computational complexity.
Test Signal Generation and Accuracy of Turntable Control in a Dummy-Head Measurement System
György Wersényi and András Illényi 150
Unlike the typical application, ultrahigh-accuracy measurements of the head-related transfer function are required when studying the influence of small changes in the acoustic environment near the head. To achieve the needed accuracy, the normal measurement method was modified. Low-crest factor, pseudo-random noise sequences replaced the conventional impulse, and the turntable positioning of the dummy head was modified for high-positional repeatability.
Loudspeaker Equalizer Design for Near-Sound-Field Applications
Wee Ser, Peng Wang, and Ming Zhang 156
Single-user loudspeaker presentation venues, such as those found in transaural and binaural systems or in personal computers, can not benefit from the conventional form of frequency response equalization. Those optimizations are on-axis signal, average power, or some other generic metric. Single-user applications benefit by optimizing the sound at the listener's two ears in the near field and no other place. The proposed method demonstrates an equalization method that improves absolute accuracy at both ears and minimizes differential errors between the two ears.
LETTERS TO THE EDITOR
Comments on "A Simplified Wavetable Matching Method Using Combinatorial Basis Spectra Selection"
Robert Bristow-Johnson 162
Author's Reply
Andrew Horner 163
STANDARDS AND INFORMATION DOCUMENTS
AES Standards Committee News 164
Analog recording; loudspeaker modeling and measurement; audio file transfer; Internet audio delivery; audio metadata
FEATURES
23rd Conference Preview, Copenhagen 170
Calendar 172
Program 174
Registration Form 179
Moving Digital Audio, Part 2-File Transfer 180
DEPARTMENTS
News of the Sections 188
Upcoming Meetings 192
Sound Track 193
New Products and Developments 195
Available Literature 197
Membership Information 198
Advertiser Internet Directory 200
Sections Contacts Directory 202
AES Conventions and Conferences 208
FPO For the cover:
2003 March, Vol 51 Number 3
spine: 2003 March, Vol 51 Number 3