### TABLE I NOTCH FILTER PARAMETERS

### Table 2. Results of rapid adaptation experiments with unsuper- vised enrollment data. RT (resp., ST) denotes the rational (resp., sine-log) all-pass transform.

"... In PAGE 4: ... Rapid Speaker Adaptation We also tested the capability of the APT to reduce the error rate of an LVCSR when used for speaker adaptation. The results of a set of experiments conducted to compare full-matrix MLLR and APT-based adaptation on a task with limited unsupervised enroll- ment data are given in Table2 ; in keeping with popular usage, we refer to this scenario as rapid adaptation. For these experiments, one global transformation was used for each speaker and CMS was applied on a per utterance basis.... ..."

### Table 2. Results of rapid adaptation experiments with unsuper- vised enrollment data. RT (resp., ST) denotes the rational (resp., sine-log) all-pass transform.

"... In PAGE 4: ... Rapid Speaker Adaptation We also tested the capability of the APT to reduce the error rate of an LVCSR when used for speaker adaptation. The results of a set of experiments conducted to compare full-matrix MLLR and APT-based adaptation on a task with limited unsupervised enroll- ment data are given in Table2 ; in keeping with popular usage, we refer to this scenario as rapid adaptation. For these experiments, one global transformation was used for each speaker and CMS was applied on a per utterance basis.... ..."

### Table 1 Filter Design

1998

"... In PAGE 2: ... The membrane filters are perforated with various shapes of holes such as circular, rectangular and hexagonal. As shown in Table1 , the opening factor of the filters is in the range of 4 % to 45 % with different hole sizes and pitches. As shown in testing section later, the pressure drop, power requirement of the filters to sustain a desired flow rate strongly depend on the opening factor.... In PAGE 3: ...5 psi while a Filter 6 (Fig. 2 (b)) in Table1 bursts at 0.9 psi.... ..."

Cited by 1

### Table 1. Comparison based on equation (6) for adaptive wavelet design of nonsubsampled wavelet transform, adaptive wavelet design of orthogonal wavelet transform and Daubechies wavelet(D12).

2002

"... In PAGE 3: ... It has been verified in our experiment that equation (6) is closely proportional to the detection performance, that is, larger value corresponds to more accurate and easier detection of fabric defects. Comparison based on this criterion for adaptive wavelet design of nonsubsampled wavelet transform, adaptive wavelet design of orthogonal wavelet transform and Daubechies wavelet is shown in Table1 . The filter length of the quadrature mirror filter for adaptive wavelet in orthogonal wavelet transform is 12, and we choose the detail space in which the value of equation (6) is larger than the other two detail spaces.... ..."

Cited by 2

### Table 5. Selected Points on Notch Transfer

2006

"... In PAGE 71: ... Figure 36. FilterPlot-Generated Notch Filter This plot and Table5 show that the response is zero at f0. At some distance away from f0, the signal is passed relatively unattenuated.... ..."

### Table 1: Algorithmic complexity in terms of required average num- ber of real multiplications and real additions per sample, and num- ber of delay elements (memory) for different realizations of a DFT FB. A real prototype filter and a real input sequence x(k) are as- sumed. The last column contains the additional operations needed for the allpass transformation of the filter-bank.

"... In PAGE 4: ... COMPARISON OF FBE AND AS FB 5.1 Algorithmic Complexity Table1 contrasts the algorithmic complexity of the derived (APT) PPN FBE to that of the (APT) PPN AS FB of Fig. 1.... In PAGE 4: ... The compu- tational complexity for the calculation of the spectral gain factors is independent of the FB and has therefore not been considered here. Table1 reveals that the uniform PPN FBE requires less sum- mations but more multiplications than the uniform PPN AS FB for most parameter configurations of L,M, and r. An important ad- vantage of the FBE is the lower number of required delay elements (memory) compared to the corresponding AS FB with the same val- ues for L, M, and r.... In PAGE 4: ... For speech enhancement with frequency warping, typical filter- bank parameters are, for example, L+1=4M=1024 and r=64 [4]. According to Table1 , the APT PPN AS FB then needs 2146 multi- plications and 5223 summations per sample, as well as 2558 delay elements. In contrast, the APT PPN FBE needs 2367 multiplica- tions and 3165 summations per sample, and 1535 delay elements.... ..."

### Table 1: Filter weights obtained by the adaptive algorithms.

"... In PAGE 19: ... The desired signal d(n) was obtained by passing s(n) through an FIR lowpass lter of window length N = 11, designed for a cuto frequency !c = 50. The weights of the designed lter are shown in Table1 , in the column entitled `Lowpass FIR apos;. Fig.... In PAGE 21: ... The step-sizes for the linear lter ( = 1:0 10?4) and the weighted median lter ( = 5:0 10?3) were chosen so that these algorithms converged in approximately the same number of iterations as the fastest weighted myriad lter algorithm (which was Algorithm II). The nal lter weights obtained by the various algorithms are shown in Table1 . The three weighted myriad lter algorithms converged to approximately the same weight vectors.... ..."

### Table 3: H2-matrix approximation with adaptive cluster bases

2003

"... In PAGE 19: ... We use a constant approximation order of m = 4 and use the algorithm from Section 5 in order to compute an orthogonalized basis. Table3 reports the results of the experiment3: The time required for building the H2-matrix approximation is roughly linear in the dimension of the discrete space, as is the time required for performing the matrix-vector multiplication. The same holds for the memory requirements: The ratio of needed storage and number of degrees of freedom is bounded.... ..."