v7.1, 20040924, me
Saturday, October 30, 9:00 am 12:30 pm
Session K SIGNAL PROCESSING, Part 1
Chair: Duane Wise, DSP Consultant, Boulder, CO, USA
K-1 New Balanced Input Integrated Circuit Achieves Very High Dynamic Range in Real-World SystemsBill Whitlock, Jensen Transformers, Van Nuys, CA, USA; Fred Floru, THAT Corporation, Milford, MA, USA
Limited Common-Mode Rejection Ratio (CMRR) in balanced interfaces often limits dynamic range in real-world audio systems. Conventional differential amplifier input circuits suffer serious CMRR degradation when driven by real system signal sources instead of laboratory generators. An ideal audio transformer, because of its extremely high common-mode impedances, is virtually immune to this degradation. A new Integrated Circuit (IC) is described that uses a patented topology to achieve common-mode impedances comparable to those of an ideal transformer. As a result, the IC enables signals with very high dynamic range to be transported without contamination by system groundvoltage differences or other sources of common-mode interference. Other features of the IC, relating to audio signal quality and reliability, are also detailed.
Convention Paper 6261
K-2 Improved Analog Class-D Amplifier with Carrier Symmetry ModulationBruce Candy, Halcro, Torrensville, Australia; S. M. Cox, University of Adelaide, Adelaide, Australia
A novel analog class-D amplifier has been developed that produces low distortion. The structure follows the well known, prior art class-D structures with negative feedback but includes modulation of the symmetry of the carrier oscillator waveform by a derivative of the input signal. This compensates a nonlinear phase modulation effect that is intrinsic to the prior art structures. The improvement is substantial at very low extra cost.
Convention Paper 6260
K-3 Enhancement of Audio Signals Using Transient Detection and ModificationMichael Goodwin, Carlos Avendano, Creative Advanced Technology Center, Scotts Valley, CA, USA
This paper describes a processing approach that enables perceptually compelling modification of audio signals via accentuation or suppression of transients. The transient detection uses a frequency-domain analysis, which yields a spectral flux parameter. In typical detection methods, such a parameter would be compared with a threshold to derive a binary transient detection function. Here, we instead use an adjustable graded response to arrive at a continuous transient characterization function. This smooth function is then used to drive a nonlinear frequency-domain signal modification. We demonstrate that binary detection is problematic for perceptual manipulation, that the soft-decision technique overcomes these problems, and that our system is able to achieve substantial modification of signals attributes without introducing significant artifacts.
Convention Paper 6255
K-4 Design Criteria for Simple Sinusoidal Parameter Estimation Based on Quadratic Interpolation of FFT Magnitude Peaks Mototsugu Abe, SONY Corporation, Tokyo, Japan; Julius O. Smith III, Stanford University, Stanford, CA, USA
Due to its simplicity and accuracy, quadratic peak interpolation in a zero-padded Fast Fourier Transform (FFT) has been widely used for sinusoidal parameter estimation in audio applications. While general criteria can guide the choice of window type, FFT length, and zero-padding factor, it is sometimes desirable in practice to know more precisely the requirements for achieving a prescribed error bound. In this paper we theoretically predict and numerically confirm the errors associated with various parameter choices and provide precise criteria for designing the estimator. In particular, we determine (1) the minimum zero-padding factor needed for a given frequency-error bound in quadratic peak interpolation, and (2) the minimum allowable frequency separation for a given window length.
Convention Paper 6256
K-5 Raised Cosine Equalization Utilizing Log Scale Filter SynthesisDavid McGrath, Justin Baird, Bruce Jackson, Surry Hills, New South Wales, Australia
An improved method of audio equalization utilizing raised cosine filters is introduced. Raised cosine filters offer improved selectivity in comparison to traditionally implemented equalization functions while also maintaining beneficial attributes such as a minimum phase response. The raised cosine filter also enables flat summation and asymmetrical filtering characteristics, resulting in an equalization system offering capability beyond traditional filter implementations.
Convention Paper 6257
K-6 Real-Time Sound Source Separation: Azimuth Discrimination and ResynthesisDan Barry, Dublin Institute of Technology, Dublin, Ireland; Bob Lawlor, National University of Ireland, Maynooth, Ireland; Eugene Coyle, Dublin Institute of Technology, Dublin, Ireland
We present a real-time sound source separation algorithm that performs the task of source separation based on the lateral displacement of a source within the stereo field. The algorithm exploits the use of the pan pot as a means to achieve image localization within stereophonic recordings. As such, only an interaural intensity difference exists between left and right channels for a single source. Gain scaling and phase cancellation techniques are used in the frequency domain to expose frequency dependent nulls across the azimuth plane. The position of these nulls in conjunction with magnitude estimation. Grouping techniques are then used to resynthesize separated sources. Results obtained from real recordings show that for music, this algorithm outperforms current source separation schemes.
Convention Paper 6258
K-7 Enhancement of Audio Signals Based on Modulation Spectrum ProcessingCarlos Avendano, Michael Goodwin, Creative Advanced Technology Center, Scotts Valley, CA, USA
In this paper we describe a signal processing technique for enhancing audio signals based on manipulation of their modulation spectra. The modification is achieved by filtering the time trajectories of spectral envelopes in different frequency bands. Scaling of higher modulation frequencies with shelving filters is used to modify rapidly-changing acoustic events, thus effectively enhancing transient components without the need for explicit detection. The perceptual effect of such modifications is analogous to the edge processing applied to images, where acoustic details can either be smoothed or sharpened depending on the desired quality of the sound.
Convention Paper 6259