Pre-filtering Technique Research Articles

Predictive coding of speech using microphone/speaker adaptation and vector quantization

A problem with speech coders based on trained quantizers is their lack of robustness against variations in the microphone and input filter response. In this paper a simple backward adaptive prefiltering technique is proposed as a means of improving the robustness and quality of a speech coder at no cost in bit rate. The technique is particularly useful in conjunction with vector quantization (VQ) of the linear predictive coding (LPC) parameters. The performance of the prefilter, denoted a microphone and speaker adaptation (MSA) filter, has been evaluated in terms of prediction gain and spectral distortion, together with objective and subjective quality of a 7.5 kbit/s CELP speech coder. In this coder a 10-bit direct VQ of the LPC parameters using the residual energy distortion measure has been applied. This is consistent with the covariance method of LPC analysis. Simulation results illustrate that the MSA filter significantly improves the performance and robustness of the LPC VQ against changes in the input response. The 7.5 kbit/s CELP with a trained excitation codebook and MSA was found to be clearly better (subjectively and objectively) than the one without MSA. The coder with MSA also showed to be practically indistinguishable from the same CELP with unquantized LPC coefficients and a stochastic excitation codebook. >

A novel structure for cascade form adaptive notch filters

The current cascade form adaptive notch filters have two different ways to adjust its parameters. First, each section is adapted individually (AI). Second, all sections are adapted simultaneously (AS). As to the complexity of tracking p sinusoids, the AI structure is linear with p but the AS structure is proportional to p 2. However, the AS structure has smaller bias in frequency estimates than that of the AI structure. In this paper, we will make a detailed comparison between the AI and AS structures by using a new prefilter technique. Then, we propose a novel structure which has the same complexity order as AIs and almost has the same lower frequency bias as ASs. Extensive computer simulations have been done to compare the performance of three structures under a wide range of tests.

The use of a cross‐correlation technique to enhance Jansson's deconvolution procedure

AbstractThe use of a cross‐correlation prefiltering technique to enhance the ability of Jansson's iterative deconvolution procedure to deconvolve extremely noisy chromatographic data is investigated. Test cases include peaks whose resolutions are as low as 0.35 and whose signal‐to‐noise ratios are as low as 1:1. Evaluation criteria include RMS error, relative peak error and peak area repeatability. For comparison purposes, relative peak area errors and peak area variances are also evaluated for noisy but well resolved peaks that have only been prefiltered with the cross‐correlation filter. Jansson's method in conjunction with cross‐correlation prefiltering is shown not only to resolve overlapped peaks but in some cases to improve their signal‐to‐noise ratios. The study also establishes some limits to the capabilities of Jansson's method will regard to adverse data conditions.

QSVD approach to on- and off-line state-space identification

Moonen et al. (1989 a), presented an SVD-based identification scheme for computing state-space models for multivariable linear time-invariant systems. In the present paper, this identification procedure is reformulated making use of the quotient singular value decomposition (QSVD). Here the input-output error co-variance matrix can be taken into account explicitly, thus extending the applicability of the identification scheme to the case were the input and output data are corrupted by coloured noise. It turns out that in practice, due to the use of various pre-filtering techniques (anti-aliasing, etc.), this latter case is most often encountered. The extended identification scheme explicitly compensates for the filter characteristics and the consistency of the identification results follows from the consistency results for the QSVD. The usefulness of this generalization is demonstrated. The development is largely inspired by recent progress in total least-squares solution techniques (Van Huffel 1989) for ...

Improved time delay estimates of underwater acoustic signals using beamforming and prefiltering techniques

The cross‐correlation method for estimating the differential time delay for an underwater acoustic signal to arrive at two spatially separated receivers is described. The output amplitudes of the two receivers are cross correlated and the lag time at which the cross‐correlation function peaks provides an estimate of the time delay. Often the time delay estimates of the signal are corrupted by the presence of noise. By replacing each of the (omnidirectional) receivers with an array of receivers and then cross correlating the beamformed outputs of the arrays, it is shown that the effect of noise on the time delay estimation process is substantially reduced. Both conventional and adaptive beamforming methods are implemented and the advantages of array beamforming (prior to cross correlation) are highlighted using both modeled data (for tutorial purposes) and experimental data. The performance of the cross correlator is further improved by using various prefiltering techniques to process the experimental data. Each of the prefilters—Hannan‐Thomson, smoothed coherence transform, and phase transform—consists of a frequency‐domain weighting function that is applied to the cross‐spectral density function before the cross‐correlation function is computed using the inverse Fourier transform. The prefilters are observed to enhance the estimation of the time delay by reducing the ambiguity associated with detecting the peak value of the basic (unweighted) cross‐correlation function, which is oscillatory in nature.

Filtering by repeated integration

Many applications of digital filtering require a space variant filter - one whose shape or size varies with position. The usual algorithm for such filters, direct convolution, is very costly for wide kernels. Image prefiltering provides an efficient alternative. We explore one prefiltering technique, repeated integration , which is a generalization of Crow's summed area table .We find that convolution of a signal with any piecewise polynomial kernel of degree n --1 can be computed by integrating the signal n times and point sampling it several times for each output sample. The use of second or higher order integration permits relatively high quality filtering. The advantage over direct convolution is that the cost of repeated integration filtering does not increase with filter width. Generalization to two-dimensional image filtering is straightforward. Implementations of the simple technique are presented in both preprocessing and stream processing styles.

Optimization of a Prefiltering Technique for the Production of Speech Highly Immune to Noise

Modification of speech by high-pass filtering and infinite amplitude clipping produces a signal having substantially greater intelligibility than that of normal speech over a wide range of signal-to-noise ratios. The cutoff frequency and attenuation slope of the high-pass filter have been varied to determine the values of these parameters which yield the highest intelligibility. The results of extensive intelligibility tests show that suppression of the first formant prior to clipping is the essential factor contributing to the higher intelligibility of modified speech. At equal signal and noise levels of 90 dB intelligibility scores greater than twice those for normal speech are consistently obtained at a number of filter settings. [Work supported by the National Aeronautics and Space Administration.]