Search

The purpose of this study was to improve the speech processing strategy for cochlear implants (CIs) based on a non-linear time-varying filter model of a biological cochlea. The level-dependent frequency response characteristic of the basilar membrane is known to produce robust formant representation and speech perception in noise. A dual resonance nonlinear (DRNL) model was adopted because it is simpler than other adaptive nonlinear models of the basilar membrane and can be readily incorporated into the CI speech processor. Spectral analysis showed that formant information is more saliently represented at the output of the proposed CI speech processor compared to the conventional strategy in noisy conditions. Acoustic simulation and hearing experiments showed that the DRNL-based nonlinear strategy improves speech performance in a speech-spectrum-shaped noise. Index Terms--Active nonlinear filter, cochlear implant (CI), dual resonance nonlinear (DRNL) model, formant representation.

A sparse structure derived for the parity-check polynomial matrices of paraunitary oversampled DFT filter banks that could be implemented at low computational cost is reported. The newly designed sparse parity-check polynomial matrices could be used to obtain paraunitary parity-check polynomials of the lowest possible order.

An iterative subspace Wiener filters for solving minimum mean-squared error (MMSE) and minimum variance unbiased estimation problems in low-dimensional subspaces is described. The results revealed that if either the conjugate gradient filter or the orthogonal multistage filter is initialized at the cross-covariance vector between the signal and the measurement then each of the subspace filters iteratively turns out a basis for a Krylov subspace.

An efficient variable fractional delay (VFD) filter is derived by using the windowing method, which has constituted an efficient alternative to the Farrow structure. The proper operation of the sampling series has helped in deriving an interpolation procedure that can be used in the design of an efficient VFD design.

A computational cyclic optimization algorithm is described for the syntheses of constant-modulus transmit signals with good auto- and cross-correlation properties. Several examples are provided for demonstrating the effectiveness of these approaches.

Optimal filters are derived on the sphere in order to detect compact objects embedded in a stochastic background process. A naive detection strategy is adopted to show the application of the new filter theory and it is shown to perform well, even at low signal-to-noise ratio.

The convex combination of filters of different families is presented for achieving an overall filter with superior tracking performance.

A two-stage genetic programming filter based on the switching scheme is presented for impulse noise filter construction. Findings reveal the capability of the proposed filter in efficiently suppressing all kinds of impulse noise.

A novel variance stabilization method is combined with undecimated wavelet, ridgelet and curvelet transforms for removing Poisson noise. Results suggest the efficiency and sensitivity of the proposed method in detecting faint features at a very low-count rate.

An efficient implementation of the nonlocal means filter based on arranging the data in a cluster tree is discussed. Findings reveal a fast and accurate preselection of similar patches.

A nonuniform subband structure with critical sampling, which is capable of modeling an arbitrary finite-impulse response (FIR) system with reduced aliasing, is described. The convergence analysis of the adaptation algorithm is used for estimating the convergence rate and steady-state mean-square error.

The effect of the residual signal energy level in sample-matrix-inversion (SMI) estimators of the optimum linear minimum-mean-square-error/linearly constrained-minimum-variance (MMSE/LCMV) filter is examined. The iterative decision-directed filter has displayed average normalized output signal-to-interference-plus-noise ratio (SINR) performance.

The estimation accuracy of the generalized adaptive notch filters (GANFs) algorithms can be considerably improved by means of compensating estimation delay effects.

A blind deconvolution method based on parametric inverse filtering is presented for medical ultrasound images. Results suggest an improvement in the resolution of ultrasound images by factors of 2.24 or 6.52 depending on the type of regularization method used.

A reduced-rank minimum description length (MDL) method is presented for the enumeration of the incident waves impinging on a uniform linear array (ULA). The proposed method is found to be better than the traditional MDL methods in detection performance and computational complexity.

The use of pitch-modulated interdigital transducers and reflectors for the realization of low-loss and wideband longitudinally coupled double-mode surface acoustic wave (DMS) filters is discussed. The pass-band becomes wide and flat, and insertion loss can be reduced.

Dual-mode filters for cellular-radio base stations require resonators with a high unloaded Q-factor, a wide spurious-free operating window and the possibility of high coupling values either to input or between resonators. These features are obtained by employing a cavity resonator loaded by grooved/splitted ceramic disks, which allows insertion of probes and/or tuning/coupling screws where the field is highly concentrated. In this paper, cavity resonator behavior, intercavity couplings, and filter computer-aided tuning are discussed in detail. Experimental results of an eight-pole transmit filter confirm the suitability of the proposed resonator for realizing cellular-radio base stations filters for UMTS and IMT-2000 applications. Index Terms--Dielectric resonators, dual-mode filters, modal techniques.

A digital cochlear filter was designed and implemented in hardware using a Xilinx XC4010 field-programmable gate array (FPGA) chip. The filter gives a good fit to biological data. It is a tenth-order recursive filter implemented as a parallel combination of low-order elements. A systematic approach to the design problem enabled us to achieve a good fit to cochlear data with efficiency of hardware usage. This approach can be easily extended to other similar applications. Index Terms--Cochlea, field-programmable gate array (FPGA), infinite-impulse response (IIR) filter.

Low-noise RF resonators and filters have been designed and constructed using either interdigital capacitor or spiral structures and made from two-inch high-temperature supeconducting (HTS) wafers. The use of HTS coils for both proton and sodium magnetic resonance imaging on humans and animals at field strengths ranging from 0.5 T to 3 T resulted in a significant increase in signal-to-noise ratio. A three-pole filter with a center frequency of 18 MHz has an insertion loss of 0.7 db at 77 K.

The use of transverse and neighboring longitudinal modes with the same symmetry to design a four-pole surface acoustic wave resonator filter was investigated. The design methodology is applicable for substrate materials with a relatively high coupling factor. Experimental results showed that the construction of both interdigital transducers arranged one after the other agrees and is characterized by a symmetrical withdrawing weighting. The design principle also permits the construction of four-pole filters without cascading or parallel connection of single filters.

The development of a surface acoustic wave resonator filter using both longitudinal and transverse modes was described. The design methodology eliminates cascading or parallel connection in the construction of four-pole filters. The input and output transducers are arranged one after the other and have equal construction characterized by unsymmetrical withdrawing weighting. The design principle also allows the suppression of neighboring disturbing longitudinal modes as a result of weighting determined by an iterative method.

A novel design methodology for a deconvolution filter that further reduces the decorrelation of the echo signal caused by tissue compression when applied in combination with signal stretching was proposed. Specifically, the correlation improvement made possible by processing the postcompression signal with a deconvolution filter following the temporal stretching stage was described. An inverse filter approach was suggested for the deconvolution, and this inverse filter was shown to be a special case of the optimal Wiener filter used in dealing with deconvolution problems.

This paper implements a filtering technique to enhance the signal-to-noise ratio (SNR) and, in turn, the detection of transient-evoked otoacoustic emissions (TEOAE's), generated by healthy human cochlea. One can increase the SNR by compiling an image of recorded TEOAE from more than one stimulus intensity, averaged over a few sweeps, which can be further processed by means of two-dimensional spatial mean filters. Averaging some 60 sweeps recorded to stimuli at several intensity levels requires one-forth of the collection time needed for a classical set of responses (average of 260 sweeps), and obtains approximately the same final SNR. The relation between the performances of the proposed technique and the SNR of the rapidly acquired responses before filtering is also investigated. Index Terms - Evoked otoacoustic emissions, fast acquisition of TEOAE, two-dimensional (2-D) filtering.

A novel integrated acoustooptic tunable filter (IAOTF) has been developed. This tunable filter is based on the Bragg interactions between waveguide and surface acoustic waves. Compared to (bulk) AOTF, its advantages include all-fiber construction, smaller size, narrower spectral resolution (1.7 nm), higher diffraction efficiency (37%), and lower rf power requirement (150 mW). A relatively narrow spectral tuning range (about 80 nm) is the only drawback for this integrated tunable filter. However, this disadvantage was overcome by judiciously using the filter for measurements in which its tuning range is coincident with the light source and also with absorption bands of analytes. In fact, an all-fiber, compact, high-throughput near-infrared spectrophotometer has been successfully constructed by synergistic use of this integrated AOTF and the erbium-doped fiber amplifier (EDFA), which has been shown to provide high intensity and wide spectral bandwidth in the near-infrared region from 1500 to 1600 nm. This spectral region is particularly useful for the determination of samples which have O - H and/or N - H groups. The all-fiber nature, compactness, high throughput, and high sensitivity of this spectrophotometer make it particularly suitable for on-line and real-time detection of trace gases in hostile environments, including leak detection of monomethylhydrazine (at a limit of detection of 191 ppm), which is often used as the hypergolic propellant for the space shuttle thruster systems.

A new structure of the subband acoustic echo canceller (AEC) is proposed. It introduces no delay in the transmission line. Using the new structure and a simplified weighted overlap-add method, an acoustic echo canceller is designed. The results showed that the designed system has good echo reduction and is stable in nonstationary environments. A hardware implementation of the proposed subband AEC is also presented in this brief.

The use of adaptive filtering fast Newton family for acoustic echo cancellation in mobile ratios reveals that low-order input signals models render the algorithm ideal for speech inputs. Numerical stabilization, tracking improvement and conditional reinitialization may also be used. The results obtained are similar to those obtained from the use of the standard LS transversal filter algorithm.

A surface transverse wave (STW) 2-port asynchronous resonator on BT-cut quartz is prepared and analyzed. STW resonators and filters on BT-cut quartz are used in applications needed for improving temperature stability.

We present a technique for digitally precompensating for shortcomings in signal reconstruction filters in an IQ modulation system. Optimum digital precompensation filters are found using least squares (LS) techniques. The optimization problem is formulated in the frequency-domain, and the optimum LS solution is characterized using an eigenanalysis approach. Index Terms: Compensation, distortion, frequency modulation, optimization methods, signal reconstruction.