Results 1 - 10 of 35,183

Table 1: Comparison of segmental SNR improvement of differ- ent one-channel noise estimation methods. PSS: Power Spec- tral Subtraction, MS: spectral subtraction based on minimum statistics [6], DLSA: log-spectral amplitude speech estimator by spectral minimum tracking [7], NSMR: the ratio of the spectral amplitude of the noisy speech to its minimum [8] and LSAHT: the proposed noise re- duction method using two stage noise filtering.

in Enhancement of Noisy Speech for Noise Robust Front-End and Speech Reconstruction at Back-End of DSR System
by Hyoung-Gook Kim, Markus Schwab, Nicolas Moreau, Thomas Sikora
"... In PAGE 3: ... Three types of background noise - white noise, car noise and factory noise - were artificially added to different portions of the data at SNR of 5 dB and -5 dB. Table1 shows that LSAHT algorithm gives best results for input SNR 5 dB and -5 dB compared to the re- sults of PSS, MS, DLSA and NSMR. Table 1: Comparison of segmental SNR improvement of differ- ent one-channel noise estimation methods.... ..."

Table 1: Comparison of segmental SNR improvement of differ- ent one-channel noise estimation methods. PSS: Power Spec- tral Subtraction, MS: spectral subtraction based on minimum statistics [1], DLSA: log-spectral amplitude speech estimator by spectral minimum tracking [7], NSMR: the ratio of the spectral amplitude of the noisy speech to its minimum [8] and LSAHT: the proposed noise reduction method using two stage noise l- tering.

in Speech Enhancement of Noisy Speech Using Log-Spectral Amplitude Estimator and Harmonic Tunneling
by Hyoung-Gook Kim, Markus Schwab, Nicolas Moreau, Thomas Sikora
"... In PAGE 3: ... Three types of background noise - white noise, car noise and factory noise - were arti cially added to different portions of the data at SNR of 5 dB and -5 dB. Table1 shows that LSAHT algorithm gives best results for input SNR 5 dB and -5 dB compared to the re- sults of PSS, MS, DLSA and NSMR.... ..."

Table 2: Parameters (ci;n;0, fi;n;0;, i;n;0) of the simulation model (RA pro le) Fig. 4 shows the ideal autocorrelation function ^ r 0(t) = J0(2 fmaxt) in comparison with the resulting autocorrelation function of the simulation model r 0(t) = r 1;0(t) + r 2;0(t) by applying the MCM and the LPNM. On the basis of the parameters listed in Table 2, we have depicted in Fig. 5 the absolute value of the fading channel amplitude (t) = j 0(t)j = q 2 1;0(t) + 2 2;0(t) ; (33) where the simulation time corresponds to a vehicle that covers in a RA environment a distance of 10m by a mobile speed of 110km/h. The probability density function (pdf) of an ideal Rayleigh process, ^ p (x), is given by [13]

in A Deterministic Method for the Derivation of a Discrete WSSUS Multipath Fading Channel Model
by Matthias Pätzold, Ulrich Killat, Yu Shi, Frank Laue

Table 4.1: Spending on Maintenance of Social Assets by the Vladimir Tractor Plantin

in Restructuring of Enterprise Social Assets in Russia: Trends, Problems, Possible Solutions
by Lev M. Freinkman, Lev M. Freinkman, Irina Starodubrovskaya, Irina Starodubrovskaya
"... In PAGE 17: ... But the attitude towards din0berent kinds of social activ- ities is the same. As one can see from Table4 , social functions associated with holding social assets are considered byenterprise managers as the most din0ecult ones. Social func- tions associated with provision of non-wage benen0cts including those in kind but which... In PAGE 18: ...transportation subsidies, direct distribution of commodities produced by the enterprise itself or received through barter at subsidized pricesn29 constitute a much less burden than the former ones. For example, Table4 shows that in general it is much more easy for en- terprises to compensate their workers for resort recreation in cash than to hold recreation facilities themselves. In addition, there is a din0berent attitude towards two din0berent groups of social assets.... In PAGE 52: ...Table4 : The Most and the Least Problematic Types of Benen0cts as Assessed byEnterprise Managers Among 3 of the Most Among 3 of the Most Din0ecult Benen0cts Easy Benen0cts to to Provide Provide Type of Benen0ct n28Number of Responsesn29 n28Number of Responsesn29 Kindergartens 15 0 Housing 10 0 Dormitories 7 0 Recreation facilities 5 0 Canteen with subsidized prices 5 3 Compensation for resort recreation 2 3 Sport facilities 2 4 Cultural center 1 0 Healthcare facilities 1 5 Food shop with subsidized prices 0 2 Transportation subsidy 0 2 Commodities on subsidized prices 0 6 Total number of responses: 24. Source: Enterprise survey.... ..."

Table 1.5 Employment in the Irish Fishing Industry, 1997 Full Time Part Time Total Full time Equivalent

in A Socio-economic Study of Fisheries in Counties Cork, Donegal, Kerry and Galway
by Prepared Gearoid, Donnchadha, Tony O Callaghan, Ciara Nil, Ionad Forbairte Duiche, Institiuid Teichneolaíochta, Trá Lí

Table 3: Reconnection probability and branching ratios for exclusive hadronic decays.

in Colour Reconnection in String Model
by Arka Todorova-Nov'a Nuclear 1996
"... In PAGE 28: ... If we now gather the information from all semileptonic decays, our apos;best guess apos; for the e ective quark masses is : mu 0:005GeV=c2 md 0:01GeV=c2 ms 0:27 0:01GeV=c2 mc 1:25 0:05GeV=c2 mb 4:3 0:1GeV=c2 These values will be used for the estimation of branching ratios in hadronic decays channels ( the expression of each amplitude is given in Appendix B). The comparison of the branching ratio obtained from the generator with the one measured experimentally is made in Table3 .( The reconnection probability for each decay channel is shown as well.... In PAGE 28: ... There is a relatively good agreement between the simulated and the experimental decay rates, which seems to corroborate the matrix elements chosen for the weak decay formalism used in this study. As seen in Table3 , a reconnection strength of about 1 is su cient to obtain the reconnection rate required by real data. Nothing seems to contradict this value as being also an acceptable scale for our studies at high energies.... ..."

Table 1: Comparison of segmental SNR improvement of differ- ent one-channel noise estimation methods. Input SNR [dB]

in Comparison of MPEG-7 Basis Projection Features and MFCC applied to Robust Speaker Recognition
by Hyoung-Gook Kim, Martin Haller, Thomas Sikora
"... In PAGE 3: ...05kHz using 16bits per sample. Table1 shows that our MLSA-MC algorithm gives best im- provement results compared to the results of MM-LSA (mul- tiplicatively modified log-spectral amplitude speech estimator) [5] and OM-LSA (optimally modified LSA speech estimator and minima controlled recursive averaging noise estimation) [6]. Table 1: Comparison of segmental SNR improvement of differ- ent one-channel noise estimation methods.... ..."

Table 2: FPGA cost-performance comparison

in Arriving at FPGA based Hardware Unix-Encription using Iterated Codesign Methods
by Ingo Cyliax, Steven D. Johnson, Bhaskar Bose, M Dcccxx
"... In PAGE 12: ... We have extracted timing estimates for a variety of Xilinx chips of di er- ent speed ratings for the XC4000E technology. Table2 shows a comparison of the various chips compared to their cost. We have run reference software implementation [12] on a variety of CPUs and list their performance and cost in Table 3.... ..."

Table 2: FPGA cost-performance comparison

in Arriving at FPGA Based Hardware Unix-Encryption Using Iterated Codesign Methods
by Ingo Cyliax, Steven D. Johnson, M Dcccx X, Ingo Cyliax 1997
"... In PAGE 12: ... Wehave extracted timing estimates for a varietyofXilinxchips of di#0Ber- ent speed ratings for the XC4000E technology. Table2 shows a comparison of the various chips compared to their cost. Wehave run reference software implementation #5B12#5D on a variety of CPUs and list their performance and cost in Table 3.... ..."

Table 1 - Results for synthetic data set with simple classification function F1.

in Using SQL primitives and parallel DB servers to speed up knowledge discovery in large relational databases.
by Alex Alves, Alex Alves Freitas, Simon H. Lavington 1996
"... In PAGE 5: ... This occurs because strong pre-pruning implies a much smaller number of queries, and these queries are associated with higher levels of the tree, where the number of examples in a tree node is larger. When the classification function is simple ( Table1 ) the time saving associated with strong pre-pruning is achieved without loss of classification accuracy. However, when the classification function is complex (Table 2) that method results in some decrease in classification accuracy.... ..."
Cited by 11
Results 1 - 10 of 35,183