AP09_2.vp 1 Introduction This paper presents some possible ways of speech en- hancement in a car cabin. This task is a very important for speech control of devices in a car or for mobile communica- tion. Both of these applications contributes to greater traffic safety. Multichannel methods of digital signal processing can be successfully used for speech enhancement. This class of meth- ods outperforms single channel methods and achieves great- er noise suppression. 2 Spatial filtering A microphone array is a basic part of multichannel pro- cessing. A uniformly spaced microphone array is the simplest arrangement. The input acoustic signal is sampled in space due to microphone spacing and in time. It is possible to dis- tinguish the signals coming from different directions thanks to spatial sampling. An input multichannel signal x[n] can be described as a mixture of the desired signal and interference. Most multi- channel systems are described under several assumptions. A model of a multichannel signal is introduced. First, the micro- phone array is focused to the Direction Of Arrival of the desired signal (DOA). Second, it is assumed that the source signal is far enough from the array. The input acoustic signal can be assumed to be a plane wave [9]. The input signal at the m-th channel can be expressed as x n s n u nm m[ ] [ ] [ ]� � , (1) where s[n] denotes the n-th sample of the desired signal, and um[n] denotes the noise and interference at the m-th sensor. 3 Interference in multichannel systems Three types of interference are usually considered in a multichannel system. A criterion for classification is the co- herence function �( )e j T� . This function expresses the re- ciprocal dependency (correlation) of particular signals in in- dividual frequency bands. The Coherence Function � ij j Te( )� of two signals is defined by the relation [14] � ij j T ij j T ii j T jj j T e e e e ( ) ( ) ( ) ( ) � � � � � � � � , (2) where � �ii j Te( ) denotes the Power Spectral Density (PSD) of a signal in the j-th channel and � �ij j Te( ) the CrossPower Spec- tral Density (CPSD) of signals in the i-th and the j-th channel. The Magnitude Squared Coherence (MSC), defined as MSC e ej T ij j T( ) ( )� �� � 2 , (3) is also often used. The type of interference is distinguished according to the shape of MSC( )e j T� . Three types of interference are recognized: spatial coherent, spatial incoherent and diffusive interference. 3.1 Spatial coherent interference First, let us consider a plane wave reaching an array of two microphones under angle �c. This situation is illustrated in Fig. 1. The spectrum of the signal at sensor 2 is X e j T2( ) � . The wavefront reaching sensor 1 is attenuated by a constant A and delayed by � �� D c c cos , (4) where c denotes the propagation speed of an acoustic signal and D denotes sensor spacing. The spectrum of the signal at sensor 1 is given by X e A X e ej T j T j T1 2( ) ( ) � � �� � . (5) Substituting (5) into (2) results in an expression for the co- herence function: 20 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ Acta Polytechnica Vol. 49 No. 2–3/2009 Noise Reduction in Car Speech V. Bolom This paper presents properties of chosen multichannel algorithms for speech enhancement in a noisy environment. These methods are suit- able for hands-free communication in a car cabin. Criteria for evaluation of these systems are also presented. The criteria consider both the level of noise suppression and the level of speech distortion. The performance of multichannel algorithms is investigated for a mixed model of speech signals and car noise and for real signals recorded in a car. Keywords: beamforming, adaptive array processing, signal processing, microphone arrays, speech enhancement. Fig. 1: An array of two sensors �12 22 2 22 22 ( ) ( ) ( ) ( ) e A e e A e e ej T j T j T j T j T j � � � � � �� � � � � � � D c ccos � . (6) Thus an expression for MSC( )e j T� reveals full coherency MSC( ) ( )e ej T j T� �� ��12 2 1. (7) 3.2 Spatial incoherent interference In case of spatial incoherent interference, the coherence computed from samples obtained at two different points in space is equal to zero in the whole frequency band, because E X e X ej T j T[ ( ) ( )]*1 2 0 � � � . X1 and X2 denote the spectra of two interferences and the asterisk denotes complex conju- gate. Incoherent noise is represented by electrical noise in microphones. 3.3 Spatial diffuse interference A reverberant environment is often encountered where many reflections occur. The delayed reflected signal reaches the array together with the direct wave. The characteristics of the delayed signal (magnitude and phase) depend on the acoustic properties of the given environment, e.g. a car cabin. This type of interference is very often present in real environ- ments, and it is called spatial diffuse interference. Diffuse noise can be modelled by an infinite number of independent sources distributed on a sphere [3]. A formula for coherence derived from this model is given by �12( ) sin e D c D c j T� � � � � � � � , (8) where � denotes angular frequency and D and c have been defined above. A shape for diffuse noise is depicted in Fig. 2. The shapes are depicted for microphone spacing D � 5cm, 10 cm and 20 cm. An analysis of equation (8) and Fig. 2 shows that the closer together the microphones are placed, the wider the main lobe of MSC is. © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 21 Acta Polytechnica Vol. 49 No. 2–3/2009 Fig. 2: MSC for diffuse noise and different microphone spacing Fig. 3: MSC for noise in a car cabin and different microphone spacing An analysis of noise recorded in a car cabin revealed inter- ference of a diffuse nature. Fig. 3 depicts the shapes of MSC for various distances of microphones. The shapes are very close to the model of diffuse noise. 4 Processing in the frequency domain Algorithms of multichannel processing can be imple- mented in the time or frequency domain. The basic algo- rithms, e.g. GSC [7], operate in the time domain. A speech signal cannot be supposed stationary, so adaptive algorithms are used. The coefficients of adaptive filters are usually con- trolled by the LMS algorithm. However, advanced algorithms require processing in the frequency domain. A block diagram of processing in the frequency domain is depicted in Fig. 4. First, the input signal is divided into quasi- -stationary overlapping segments. Moreover, each segment is weighted by a Hamming window. A typical segment length is 16 ms. Second, a short time spectrum is computed. Third, the short-time spectra are processed. An input signal is finally ob- tained using the inverse Fourier transform and the Overlap and Add (OLA) method [14]. Weight adaptation is performed block by block. The adap- tation is performed according to Minimum Mean Square Error (MMSE). The advantage of this approach is that the weights in each frequency band change according to the power of the noise in a particular band. 5 Beamforming algorithms The performance of four algorithms will be presented in this paper. Their principles will be explained in this section. The following algorithms will be presented: Beamformer with Adaptive Postprocessing (BAP) [16], Generalised Sidelobe Canceler (GSC) [7], Linearly Constrained Beamformer (LCB) [5] and Modified Coherence Filtering (MCF) [10]. 5.1 BAP Delay And Sum beamformer (DAS) is the first block of BAP [16]. The output of this block Yb is an average of the input channels. Weights wi are equal to 1 M. BAP improves the DAS beamformer by using a Wiener Filter (WF) behind the DAS structure, Fig. 5. The main contribution of WF is in improving the suppression level of uncorrelated inter- ferences. The derivation for the weights of WF can be found in [15]. Weights in the frequency domain are obtained as W e e e j T xs j T xx j T ( ) ( ) ( ) � � � � � � , (9) where � �xx j Te( ) denotes the Power Spectral Density (PSD) of the signal x[k] (input of WF), and � �xs j Te( ) is the Cross- -Power Spectral Density (CPSD) of the signals x[k] and s[k] (output of WF). It is assumed that the interferences are uncorrelated ( [ ( ) ( )]E U e U ei j T j j T� � � 0 for all i j� ) and the desired signal is uncorrelated with the interferences ( [ ( ) ( )]E S e U ej T j j T� � � 0 for all i). S e j T( )� ) is a spectrum of the desired signal and U ei j T( )� is a spectrum of the interfer- ence at the i-th sensor. Under these assumptions it holds � � � � � � xs j T sx j T ss j Te e e( ) ( ) ( )� � . (10) Weights of WF can now be expressed as W e e e j T ss j T xx j T ( ) ( ) ( ) � � � � � � (11) In the case of the BAP structure, the PSDs in relation (11) are estimated by averaging the signal in a particular channel [13] � � ( ) Re ( ) ( )*� � �ss i j T k j T k i M i M M M X e X e� � � �� � ��2 1 11 1 , (12) � ( ) ( ) ,� � �xx j T i j T i M e M X e� � �1 1 2 (13) where X ei j T( )� is a spectrum of the input signal. 5.2 GSC The Structure of GSC [7] is depicted in Fig. 6. It is equal to the Adaptive Beamformer [6]. The system consists of the DAS beamformer and the Adaptive Noise Canceler (ANC). ANC serves to suppress the coherent interference. The weights of ANC filters are in accordance with Wiener theory [7]. A formula for optimal weights is given by H e e e i j T Y Y j T Y Y j T i b i i ( ) ( ) ( ) � � � � � � , i M� �1 1, ,� . (14) � � Y Y j T i b e( ) denotes the CPSD of signals Yi and Yw, the mean- ing of which is obvious from Fig. 6. � �Y Y j T i i e( ) is the PSD of Yi. The Proper function of the ANC is given by perfect sepa- ration of the desired signal from the input signal. Let us denote any coherent signal incident on the array from any 22 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ Acta Polytechnica Vol. 49 No. 2–3/2009 Fig. 4: Block diagram of processing in the frequency domain Fig. 5: BAP direction except DOA as coherent interference. Under this assumption, an interference can be separated from the input signal by an appropriate combination of input channels xi[k]. This separation is arranged by the Blocking Matrix (BM). The most commonly used BM differentiates neighbouring channels. BM consists of M columns, and (M � 1) rows, and looks like this [7]: BM � � � � � � � � � � � � 1 1 0 0 0 0 1 1 0 0 0 0 0 1 1 � � � � � � � � � . (15) 5.3 LCB LCB utilizes GSC and BAP beamformers [5]. The struc- ture of LCB is depicted in Fig. 7. The direct branch composed of BAP suppresses inco- herent interference. The lower branch consisting of ANC is responsible for coherent interference suppression. The greatest difference between GSC and LCB is the way in which the weights of the ANC filters are computed. In LCB they are computed from signals at the outputs of BM and WF. The relation for calculating ANC filters has to be written as H e e e i j T Y Y j T Y Y j T i w i i ( ) ( ) ( ) � � � � � � , i M� �1 1, ,� . (16) � Y Yi w denotes the CPSD of signals Yi and Yw the meaning of which is obvious from Fig. 7. 5.4 Coherence Filtering Coherence Filtering differs from the other multichannel systems. It is a representative of double channel methods. The idea of this method [2] is based on the fact that the coherence function of the spatially coherent desired signal is close to one, and the coherence of the incoherent interference is close to zero. The authors of [10] propose a modification to Coherence Filtering. The Coherence Filter is included in the BAP struc- ture, see Fig. 8. The coefficients of the Modified Coherence Filter (MCF) C(k) are computed as follows C k k T k k T ( ) , � � � � � � W(k), if ( ) ( ) if ( ) � � � � (17) where W(k) denotes an estimated frequency response of the Wiener filter, equation (9), and T denotes the threshold. © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 23 Acta Polytechnica Vol. 49 No. 2–3/2009 Fig. 6: GSC Fig. 7: LCB Fig. 8: Structure of a Modified Coherence Filter 6 Testing procedure It is very difficult to separate the desired signal and noise when the level of noise suppression is evaluated. Separation of the desired signal and noise is crucial for an assessment of the properties of the algorithms. The following approach has been chosen for testing the algorithms. The desired signal and interference are recorded separately. The input mixture x [n] is obtained before processing, so that the SNR is defined. The recordings of utterances in a standing car with the engine switched off are assumed to be the desired signal. Noise is rep- resented by recordings of noise in a moving car without the presence of speech. A block diagram of the testing procedure is depicted in Fig. 9. The input signals s [n] (desired signal) and u[n] (noise) are mixed with defined SNR to make x [n]. The output signal y[n] originates by processing the input mixture. During processing, the coefficients of the adap- tive filters are set. Using these coefficients, clear signals s [n] and u[n] are also processed. This processing results in out- put signals ys[n] and yu[n]. These signals carry information about the influence of the system on the desired signal and interference. 7 Criteria for system evaluation The criteria for assessing the level of speech enhancement can be classified into two classes, objective and subjective. The subjective criteria are represented by listening tests. Listening tests are very difficult. It is necessary to gather several quali- fied listeners. The test also consumes a great deal of time. However, these tests can show how the output signals are per- ceived by human subjects. Objective criteria give exact infor- mation and are not influenced by external factors, e.g. the mood of the listener. The following criteria will be used for evaluating the algorithms: Noise Reduction (NR), Log Area Ratio (LAR), Signal to Noise Ratio Enhancement (SNRE) and spectrograms. All of the criteria will be computed from quasi-stationary segments of the signal. 7.1 Noise reduction NR expresses the ability of an algorithm to reduce noise. NR is defined as NR( ) ( ) ( ) e e e j T uu j T y y j T u u � � � � � � , (18) where �uu j Te( )� is the PSD of the interference at the input of the system, and �y y j T u u e( )� is the PSD of the interference processed by the system. The assumption for NR calculation is that no desired signal is present at the input of the system. NR considers only the influence of the system on the inter- ference. It does not consider the influence on the desired signal. This criterion has to be combined with other criteria. 7.2 Log Area Ratio LAR [12] takes into account the influence of the system on the desired signal and speech intelligibility. An advantage of this criterion is its high correlation with listening tests [4]. A presumption when using this criterion is the presence of speech. LAR is calculated on the basis of the partial correla- tion coefficients (PARCOR) of the auto regressive model [8]. Computing LAR requires a clear speech signal s[n] and an output signal ys[n]. The computing is performed in the fol- lowing steps: 1. Estimation of PARCOR coefficients k (p, l) of the signal segment. Index p denotes the p-th PARCOR coefficient and l the signal segment. The order of the model is cho- sen as P � 12. A Burg algorithm can be used for estimat- ing the coefficients [8]. 2. Calculation of area coefficients g p l k p l k p l p( , ) ( , ) ( , ) , , , ,� � � � 1 1 1 12� (19) where k (p, l) is a p-th PARCOR coefficient of the l-th seg- ment. (PARCOR coefficients k (p, l) are marked in some sources [11] as a negative of reflection coefficients.) 3. Calculation of LAR for block l LAR ( ) log ( , ) ( , ) l g p l g p l s yp � � �20 10 1 12 . (20) LAR expresses the “distance” of the model of signal s [n] from the model of signal y[n]. The lower LAR is, the less the speech is distorted. 7.3 SNRE SNRE is very often used for evaluating systems for speech enhancement. The value of SNRE is also calculated segment by segment. SNRE is obtained as the difference of SNRout – SNRin. Signals s [n] and u[n] are used for calculating SNRin and ys[n] and yu[n] are user for calculating SNRout. 24 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ Acta Polytechnica Vol. 49 No. 2–3/2009 Fig. 9: Block diagram of the testing procedure 8 Database of car speech A database of car speech and car noise has been created for developing and verifying of multichannel systems per- forming speech enhancement. The database creation proce- dure was chosen to fulfill the requirements of the testing procedure described in section 6. The signals were recorded in a Škoda Fabia. A microphone array of 12 sensors with constant spacing of 4 cm was used. The desired signal was represented by reproduced recordings of female and male utterances. Noise signals were recorded under various condi- tions. More details about the database are summarized in [1]. 9 Experiments Two approaches were used to verify the algorithms pre- sented in this paper. First, a model of the desired signal and noise recorded in a car were used as an input mixture. A model of the desired signal was created by copying the clear speech signal into all channels. The purpose of this approach is to verify the performance of the algorithm. The influence of the properties of the microphone array is not considered. Breaking the assumptions mentioned in section 3 introduces additional delays of signals between the individual mi- crophones. Additional delays can be due to the fact that the acoustic signals cannot be represented by plane waves, and due to array imperfections. Solving these problems is a separate issue. The purpose of the second experiment is to show the properties of the whole system. It should show that the prop- erties of the array are significant and that it is worth taking them into account. Each of the experiments was performed for two different environments. The first environment was a standing car with a running engine, and the second environment was a car moving outside a village (70 km/h). The criteria NR, LAR and SNRE were calculated for segments of 128 samples. An mean value was calculated for each criterion. The experiments were performed for an array of 4 mi- crophones with 4 cm spacing and SNRin was set to 0 dB. The sample rate was 8 kHz. The parameters of MCF were set to T � 0 2. and � � 2. Tables 1 and 2 show the results for a model of the desired signal. The results for a real signal are displayed in Tables 3 and 4. The experiment with a model of the signal was done for different values of SNRin. The results are summarized in the graphs in Figs. 10, 11 and 12. 10 Conclusion The experiments enabled a comparison of the methods for speech enhancement presented here. The results are very different for a model of the desired signal and for a real sig- nal. Array imperfections and propagation of signals are the most important influences. LCB provided the best results for a model of a signal. The experiments showed the importance of using several criteria. BAP achieves low SNRE and high NR according to the results in Table 1. GSC behaves in an opposite way. MCF seems to have the weakest performance. It produced high speech dis- tortion (high values of LAR) and low SNRE and NR. Zero speech distortion is worth noting in the case of GSC. This is due to perfect separation of the desired signal at the input of the ANC filters. © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 25 Acta Polytechnica Vol. 49 No. 2–3/2009 LAR SNRE NR BAP 0.53 0.9 13.85 GSC 0.0 3.11 �1.46 LCB 0.53 3.22 10.86 MCF 1.83 1.38 2.72 Table 2: Results for a model of a signal, running car (70 km/h) LAR SNRE NR BAP 4.42 �0.32 13.66 GSC 7.36 �1.8 4.40 LCB 7.62 �1.9 17.24 MCF 6.26 �0.33 2.84 Table 3: Results for a real signal, standing car LAR SNRE NR BAP 4.41 �0.78 14.64 GSC 7.25 �1.47 5.95 LCB 7.52 �1.51 18.95 MCF 6.38 �0.69 4.21 Table 4: Results for a real signal, running car (70 km/h) LAR SNRE NR BAP 0.5 1.52 13.37 GSC 0.0 4.95 2.43 LCB 0.5 5.14 14.75 MCF 1.66 2.19 2.33 Table 1: Results for a model of a signal, standing car 26 © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ Acta Polytechnica Vol. 49 No. 2–3/2009 Fig. 10: NR for various SNRin Fig. 11: LAR for various SNRin Fig. 12: SNRE for various SNRin All of the algorithms showed much worse performance for real signals. There is both high speech distortion and low enhancement. There are no so significant differences between a standing car and a moving car. NR for BAP and LCB is an exception. The lowest values of LAR and SNRE are for BAP and MCF. The second experiment focused on the influence of SNRin on the results. The shape of NR (Fig. 10) reveals strong de- pendance on SNRin for GSC and LCB. The NR of LCB falls below BAP, and GSC falls below MCF for high values of SNRin. The shape of SNRE (Fig. 12) shows a very similar trend. BAP and MCF are almost independent of SNRin with respect to both NR and SNRE. Only BAP, LCB and MCF can be considered when ob- serving LAR (Fig. 11). GSC does not distort speech in the case of a model of the input signal, due to perfect separation of the desired signal. BAP and LCB have the same shape of LAR for the same reason. The highest speech distortion was for MCF. The figure also shows that speech distortion decreases with growing SNRin. This paper has shown the properties of selected algo- rithms for speech enhancement in a noisy environment. The experiments with a model of the input signal showed that these methods are capable of speech enhancement. A prob- lem occurred when the methods were used for real signals. The assumptions of proper functionality were broken in this case. The input signals did not match the model that the methods were developed for. It is necessary to focus on com- pensating the array imperfections and signal propagation in future work. Acknowledgments The research described in this paper was supervised by prof. Pavel Sovka. This paper was mainly supported by research activity MSM 6840770012 ”Transdisciplinary Research in Biomedical Engineering II” and GAČR grant 102/08/H008 “Analysis and modelling of biomedical and speech signals” and GAČR grant GA102/08/0707 “Speech Recognition under Real-World Conditions”. References [1] Bolom, V., Sovka, P.: Multichannel Database of Car Speech. In Digital Technologies 2008, Vol. 1 (2008), Žilina: University of Žilina, Faculty of Electrical Engineering. [2] Le Bouquin, R.: Enhancement of Noisy Speech Signals: Application to Mobile Radio Communications. Speech Commun., Vol. 18 (1996), No. 1, p. 3–19. [3] Cron, B. F., Sherman, C. H.: Spatial-Correlation Func- tions for Various Noise Models. Journal of Acoustic Society of America, Vol. 34 (1962), No. 1. [4] Fischer, S., Kammeyer, K.-D., Simmer, K.U.: Adaptive Microphone Arrays for Speech Enhancement in Coher- ent and Incoherent Noise Fields. In Invited talk at the 3rd joint meeting of the Acoustical Society of America and the Acous- tical Society of Japan, Honolulu, Hawaii, December 1996. [5] Fischer, S., Simmer, K. U.: Beamforming Microphone Arrays for Speech Acquisition in Noisy Environments. Speech communication, Vol. 20 (1996), p. 215–227. [6] Frost, O. L.: An Algorithm for Linearly Constrained Adaptive Array Processing. In Proceedings of the IEEE, Vol. 60 (1972), p. 926–934. [7] Griffiths, L. J., Jim, W. C.: An Alternative Approach to Linearly Constrained Adaptive Beamforming. Antennas and Propagation, IEEE Transactions on, Vol, 30 (1982), p. 27–34. [8] Haykin, S.: Adaptive Filter Theory (3rd ed.), Upper Saddle River, NJ, (USA): Prentice-Hall, Inc., 1996. [9] Herbordt, W.: Sound Capture for Human/Machine Inter- faces. Practical Aspects of Microphone Array Signal Processing. Springer, 2005. [10] Mahmoudi, D., Drygajlo, A.: Combined Wiener and Coherence Filtering in Wavelet Domain for Micro- phone Array Speech Enhancement. Acoustics, Speech and Signal Processing, 1998. Proceedings of the 1998 IEEE Inter- national Conference on, 12–15 May 1998, Vol. 1 (1998), p. 385–388. [11] Psutka, J., Müller, L., Matoušek, J., Radová, V.: Mluvíme s počítačem česky. Prague: Academia, 2006. [12] Simmer, K.U., Bitzer, J., Marro, C.: Post-filtering Tech- niques. In Microphone Arrays, Berlin, Heidelberg, New York: Springer, May 2001, p. 39–57. [13] immer, K.U., Wasiljeff, A.: Adaptive Microphone Arrays for Noise Suppression in the Frequency Domain. In COST-229 Workshop on Adaptive Algorithms in Communica- tions, Bordeaux, France, Sep 1992, p. 185–194. [14] Uhlíř, J., Sovka, P.: Číslicové zpracování signálů. Prague: CTU – Publishing House, 2002. [15] Widrow, B., Stearns, S. D.: Adaptive Signal Processing. Prentice-Hall, 1985. [16] Zelinski, R.: A Microphone Array with Adaptive Post-Fil- tering for Noise Reduction in Reverberant Rooms. In In- ternational Conference on Acoustic Speech Signal Processing, New York, 1988, p. 2578–2581. Václav Bolom e-mail: bolomv1@fel.cvut.cz Department of Circuit Theory Czech Technical University in Prague Faculty of Electrical Engineering Technická 2 166 27 Prague, Czech Republic © Czech Technical University Publishing House http://ctn.cvut.cz/ap/ 27 Acta Polytechnica Vol. 49 No. 2–3/2009 << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (Dot Gain 20%) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Error /CompatibilityLevel 1.4 /CompressObjects /Tags /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /CMYK /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams false /MaxSubsetPct 100 /Optimize true /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo true /PreserveFlatness true /PreserveHalftoneInfo false /PreserveOPIComments true /PreserveOverprintSettings true /StartPage 1 /SubsetFonts true /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 300 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.50000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages true /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /ColorImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 300 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 1200 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description << /ARA /BGR /CHS /CHT /CZE /DAN /DEU /ESP /ETI /FRA /GRE /HEB /HRV (Za stvaranje Adobe PDF dokumenata najpogodnijih za visokokvalitetni ispis prije tiskanja koristite ove postavke. Stvoreni PDF dokumenti mogu se otvoriti Acrobat i Adobe Reader 5.0 i kasnijim verzijama.) /HUN /ITA /JPN /KOR /LTH /LVI /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken die zijn geoptimaliseerd voor prepress-afdrukken van hoge kwaliteit. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.) /NOR /POL /PTB /RUM /RUS /SKY /SLV /SUO /SVE /TUR /UKR /ENU (Use these settings to create Adobe PDF documents best suited for high-quality prepress printing. Created PDF documents can be opened with Acrobat and Adobe Reader 5.0 and later.) >> /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ << /AsReaderSpreads false /CropImagesToFrames true /ErrorControl /WarnAndContinue /FlattenerIgnoreSpreadOverrides false /IncludeGuidesGrids false /IncludeNonPrinting false /IncludeSlug false /Namespace [ (Adobe) (InDesign) (4.0) ] /OmitPlacedBitmaps false /OmitPlacedEPS false /OmitPlacedPDF false /SimulateOverprint /Legacy >> << /AddBleedMarks false /AddColorBars false /AddCropMarks false /AddPageInfo false /AddRegMarks false /ConvertColors /ConvertToCMYK /DestinationProfileName () /DestinationProfileSelector /DocumentCMYK /Downsample16BitImages true /FlattenerPreset << /PresetSelector /MediumResolution >> /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles false /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /DocumentCMYK /PreserveEditing true /UntaggedCMYKHandling /LeaveUntagged /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ] >> setdistillerparams << /HWResolution [2400 2400] /PageSize [612.000 792.000] >> setpagedevice