Subject: AT&T TrueVoice DSP Analysis Date: Fri, 3 Sep 93 13:39:17 -0700 From: malcolm@apple.com I've just finished an analysis of the AT&T TrueVoice demo that has been advertised by AT&T and discussed in the TELECOM Digest. Here are the results of the analysis. First, there is a reason that the TrueVoice version sounds louder. Above 300Hz there is a pretty uniform gain of 4dB. Between 100-300 Hz there is a 14 dB gain peak (with a notch at 180 Hz to kill the third harmonic of the power line hum.) Here are the details: 0-100 Hz Not much change. 100-300Hz A broad gain bump of about 14dB 180Hz Deep 12.5dB notch (back to 1.5 dB) 300-4kHz 4dB extra gain with a .5dB ripple Method: The AT&T TrueVoice demo (1 800 932-2000) was digitized with the Telecom pod on the Macintosh Quadra 840AV (using the AT&T 3210 DSP chip). Samples were converted with 16 bits of linear resolution, sampled at 8kHz, and stored in an AIFF file. All data was collected from a single phone call using an analog phone line. The results were loaded into Matlab on a Macintosh for analysis. The TrueVoice demo starts with 30 seconds of normal speech and then automatically switches on the TrueVoice modification and replays the same speech. The beginning of each demo sample was aligned by eye, and then a cross correlation was used to refine the relative starting positions. All further analysis was done on these aligned samples of modified and unmodified speech. Note, the TrueVoice modification is turned on over a second or two. (I don't know why AT&T does this. Perhaps they didn't want the change to seem too jarring.) The cross correlation analysis was done using these first couple of seconds: this should make the correlation more accurate. The spectral analysis that followed skipped the first three seconds. A periodogram analysis was performed using a window of data of length 2048 and skipping 16 samples between windows. A Hamming window was used before FFT'ing and the power spectral density was computed. The power spectral density for all windows was summed (averaged) to arrive at a robust estimate of the power spectral density of the entire sample. A total of 170,000 data samples was used from each example. The power spectral density for each example (unmodified and modified speech) was converted to dB. The two PSDs are then subtracted to estimate the difference in spectral energy due to the TrueVoice filtering. Results: The broad filter characteristics are shown below. 0-100 Hz Not much change. 100-300Hz A broad gain of 13.7 dB, 3dB Q is 1.5 120Hz About a 1dB dip in the gain 180Hz Narrow band reject down to 1.5 dB, 3dB Q is 18 300-4kHz 4dB extra gain with a .5dB ripple Note, there is a deep notch at the 3rd harmonic of the power line frequency and a extra small dip at the second harmonic. I don't have enough information to tell if they are doing anything different at 60Hz. The filtering appears to be stationary. The demo over the phone turns the filtering on gradually, but after a second there appears to be no change in the filter's characteristics. The analysis described above was performed over several subsections of the data and there was no visible change in the filter's characteristics. Acknowledgements: Many thanks to Stuart Davidson for setting up the hardware and software to digitize the demo. Thanks also to Michele Covell (Interval Research) for help with the filter analysis. The periodogram is discussed in Larry Marple's book "Digital Spectral Analysis with Applications." Disclaimer: I have no affiliation with AT&T except as a mostly satisfied customer. This work is not an official Apple project. TrueVoice and AT&T are undoubtably trademarks of AT&T. Malcolm Slaney malcolm@apple.com =================== Matlab Periodogram Code ============================ % function y=periodogram(x,width,skip) % Compute the periodogram (an estimate of the power spectral % density) of an array. The width of each analysis defaults to % 256 points, and the default skip between windows is half the % window size. function y=periodogram(x,width,skip) if (nargin < 2) width = 256; end if (nargin < 3) skip = width/2; end nwinds = fix((length(x)-width)/skip); x = reshape(x,length(x),1); wind = hamming(width); y = zeros(width,1); for i=1:nwinds firstPoint = (i-1)*skip + 1; d = [x(firstPoint:firstPoint+width-1).*wind;zeros(width,1)]; f = fft(d); f = f(1:width); y = y + f .* conj(f); end y = y/nwinds; =============== Matlab TrueVoice Analysis Code ========================= % Analyze a recorded sample of the AT&T TrueVoice demo. % This script assumes that the entire demo is stored in % an array called tv. The start of the first and second % version of the demo are indicated by the first and second % pointers. (These will be different for anybody else's % recording of the phone call.) first=116000; second=395500+41; firstEnd = 355000; secondEnd = 610000; sampleRate = 8000; %segment = 170000; segment = 32768; offset = 30000; if (second + offset + segment > secondEnd) disp('Segment is past end of second example.'); return; end s1=tv(first+offset:first+offset+segment-1); s2=tv(second+offset:second+offset+segment-1); % Optionally compute the optimum cross correlation. % If the alignment is done right, then the maximum % value of the cross correlation should be in the % middle of the cross correlation or sample number % equal to segment size . % cor = xcorr(s1,s2); % plot(cor) % [m,i] = max(cor); % i r1 = 10*log10(periodogram(s1, 512, 128)); r2 = 10*log10(periodogram(s2, 512, 128)); f = (0:length(r1)-1)/length(r1)*sampleRate/2; =============== Spectral Analysis of TrueVoice ========================= % Spectral analysis (in dB) of the AT&T TrueVoice demo. % The results are shown below at 1.9531Hz intervals from % 0Hz to nearly 4kHz. -0.0156 -0.0156 -0.0154 -0.0138 2.8762 1.5637 1.3204 0.2571 0.6921 0.9584 0.7187 -0.7438 -1.7550 -2.0471 -2.0754 -1.9768 -1.6302 -0.7998 0.7231 0.9736 0.9899 0.9882 0.9827 0.9418 0.2723 -0.3172 -0.4245 0.0216 -0.1162 -0.1018 -0.0875 -0.0747 -0.0601 -0.0360 0.1168 0.9137 2.4710 2.5066 3.5501 2.9577 3.3023 3.0256 3.8413 4.5084 6.3061 6.4052 7.6116 7.6745 8.5071 8.4967 9.0233 9.0818 9.5900 9.7797 10.1627 10.3327 10.6086 10.7367 10.9011 10.8782 10.5145 10.0140 10.3593 11.2615 11.8157 11.9453 12.0295 12.0388 12.0603 12.1355 12.3434 12.4985 12.6426 12.7032 12.8563 12.9541 13.0520 13.0989 13.2080 13.2716 13.3565 13.3807 13.4805 13.5283 13.5936 13.6157 13.6703 13.6777 13.6661 12.1029 5.5750 2.3909 1.6613 2.2414 4.9604 11.2490 13.7420 13.7310 13.7230 13.6977 13.6595 13.5861 13.5226 13.4833 13.4672 13.4355 13.3998 13.3534 13.3033 13.2455 13.1833 13.1151 13.0436 12.9571 12.8699 12.7882 12.6880 12.5852 12.5082 12.4377 12.3462 12.2276 12.1066 11.9765 11.8630 11.7580 11.6211 11.4772 11.3559 11.2418 11.0955 10.9580 10.8388 10.6905 10.5217 10.3533 10.2052 10.0873 9.9788 9.8533 9.7000 9.5401 9.3863 9.2267 9.0551 8.8746 8.7166 8.5856 8.4393 8.2574 8.0735 7.8216 7.2432 6.4982 6.2854 6.7300 7.0660 7.0194 6.8627 6.6837 6.4944 6.3324 6.1924 6.0580 5.9400 5.8304 5.6939 5.5120 5.3422 5.2091 5.0764 4.9489 4.8327 4.7075 4.5998 4.4920 4.3769 4.2938 4.2189 4.1290 4.0337 3.9499 3.8720 3.7967 3.7345 3.6756 3.6097 3.5443 3.4910 3.4470 3.4026 3.3690 3.3538 3.3306 3.2824 3.2459 3.2375 3.2455 3.2562 3.2654 3.2769 3.2867 3.2941 3.3026 3.3110 3.3201 3.3315 3.3440 3.3697 3.4085 3.4461 3.4825 3.5228 3.5549 3.5623 3.5717 3.6202 3.6865 3.7361 3.7728 3.7982 3.8225 3.8617 3.8934 3.9120 3.9411 3.9805 4.0103 4.0313 4.0423 4.0499 4.0719 4.1125 4.1634 4.1984 4.2129 4.2306 4.2533 4.2706 4.2897 4.3161 4.3271 4.3166 4.3097 4.3212 4.3222 4.3041 4.3011 4.3121 4.3238 4.3343 4.3339 4.3267 4.3170 4.2965 4.2704 4.2473 4.2342 4.2294 4.2322 4.2381 4.2338 4.2211 4.2096 4.2020 4.1905 4.1708 4.1571 4.1513 4.1370 4.1146 4.0957 4.0799 4.0686 4.0589 4.0159 3.9430 3.9157 3.9485 3.9754 3.9758 3.9720 3.9625 3.9376 3.9162 3.9101 3.9068 3.9034 3.9000 3.8920 3.8796 3.8663 3.8585 3.8588 3.8562 3.8374 3.8205 3.8194 3.8240 3.8239 3.8113 3.8002 3.8025 3.8091 3.8143 3.8212 3.8278 3.8367 3.8441 3.8418 3.8338 3.8264 3.8311 3.8473 3.8638 3.8787 3.8898 3.8965 3.8994 3.8999 3.9121 3.9445 3.9715 3.9803 3.9848 3.9975 4.0300 4.0597 4.0501 4.0212 4.0095 4.0264 4.0668 4.1108 4.1344 4.1324 4.1079 4.0761 4.0591 4.0615 4.0805 4.1049 4.1293 4.1566 4.1745 4.1746 4.1766 4.1942 4.2163 4.2459 4.2697 4.2636 4.2516 4.2672 4.2902 4.2960 4.2897 4.2854 4.2939 4.3081 4.3115 4.3180 4.3467 4.3717 4.3674 4.3452 4.3170 4.2937 4.2886 4.2898 4.2866 4.2869 4.2744 4.2453 4.2252 4.2041 4.1907 4.1994 4.2001 4.1973 4.2036 4.2054 4.1940 4.1572 4.1137 4.1197 4.1507 4.1396 4.1036 4.0570 4.0095 3.9778 3.9774 3.9987 4.0189 4.0231 4.0033 3.9736 3.9444 3.9261 3.9289 3.9371 3.9147 3.8360 3.7424 3.7082 3.7061 3.6903 3.6718 3.6725 3.6802 3.6756 3.6680 3.6843 3.7143 3.7304 3.7240 3.7022 3.6766 3.6636 3.6838 3.7364 3.7735 3.7715 3.7745 3.8092 3.8338 3.8216 3.8117 3.8196 3.8251 3.8316 3.8483 3.8663 3.8788 3.8834 3.8788 3.8763 3.8858 3.8962 3.8959 3.8940 3.9102 3.9491 3.9801 3.9850 4.0079 4.0531 4.0789 4.0709 4.0647 4.0794 4.0913 4.0958 4.1114 4.1341 4.1524 4.1771 4.1970 4.2059 4.2127 4.2154 4.2141 4.2150 4.2213 4.2226 4.2111 4.1941 4.1933 4.2252 4.2576 4.2572 4.2150 4.1531 4.1139 4.1130 4.1501 4.1874 4.1887 4.1605 4.1311 4.1236 4.1473 4.1852 4.2052 4.2233 4.2426 4.2489 4.2538 4.2531 4.2337 4.2260 4.2384 4.2403 4.2367 4.2453 4.2668 4.2789 4.2627 4.2359 4.2013 4.1635 4.1479 4.1810 4.2237 4.2029 4.1435 4.1030 4.0890 4.0912 4.1109 4.1343 4.1405 4.1113 4.0482 3.9993 3.9791 3.9588 3.9272 3.8873 3.8515 3.8288 3.8098 3.8065 3.8195 3.8113 3.7923 3.8059 3.8273 3.8364 3.8258 3.7813 3.7552 3.7867 3.8330 3.8302 3.8012 3.7939 3.7789 3.7384 3.7168 3.7019 3.6732 3.6598 3.6861 3.7230 3.7378 3.7430 3.7643 3.7819 3.7818 3.7938 3.7803 3.7341 3.7190 3.7303 3.7457 3.7783 3.8146 3.8645 3.9324 3.9773 4.0000 4.0181 3.9884 3.9119 3.8909 3.9387 4.0071 4.0808 4.1228 4.1229 4.1029 4.0704 4.0309 4.0110 4.0390 4.0767 4.0965 4.1365 4.1915 4.2151 4.2345 4.2727 4.2879 4.2576 4.2202 4.2258 4.2612 4.2724 4.2555 4.2427 4.2471 4.2661 4.2896 4.3010 4.3005 4.2957 4.2855 4.2740 4.2666 4.2695 4.2819 4.2873 4.2675 4.2188 4.1749 4.1607 4.1578 4.1489 4.1283 4.1087 4.1226 4.1646 4.1975 4.2025 4.1791 4.1383 4.1318 4.1720 4.1981 4.1785 4.1589 4.1588 4.1699 4.1828 4.1738 4.1261 4.0842 4.0948 4.1499 4.1747 4.1536 4.1222 4.0861 4.0489 4.0195 4.0093 4.0231 4.0133 3.9700 3.9524 3.9736 4.0181 4.0664 4.1272 4.1630 4.1220 4.0943 4.0931 4.0471 3.9736 3.9507 3.9191 3.8515 3.8284 3.8700 3.9406 3.9809 3.9753 3.9601 3.9425 3.9219 3.9120 3.8903 3.8508 3.8170 3.7936 3.7934 3.8167 3.8262 3.8181 3.8137 3.7817 3.7410 3.7353 3.7615 3.8004 3.8142 3.8099 3.8200 3.8412 3.8395 3.8060 3.7680 3.7490 3.7546 3.7850 3.8190 3.8374 3.8329 3.8101 3.7999 3.8239 3.8626 3.8995 3.9293 3.9370 3.9439 3.9866 4.0243 4.0340 4.0458 4.0547 4.0425 4.0240 4.0253 4.0536 4.0953 4.1350 4.1705 4.1965 4.2225 4.2664 4.2941 4.2819 4.2690 4.2711 4.2839 4.2996 4.3138 4.3118 4.2843 4.2663 4.2734 4.2854 4.2986 4.3212 4.3424 4.3508 4.3505 4.3336 4.3145 4.3226 4.3632 4.4118 4.4020 4.3286 4.2562 4.2137 4.2224 4.2605 4.2589 4.2231 4.2219 4.2602 4.2963 4.2905 4.2564 4.2230 4.2089 4.2149 4.2181 4.2004 4.1570 4.0962 4.0639 4.0803 4.0981 4.1046 4.0952 4.0567 4.0047 3.9813 3.9860 3.9971 4.0172 4.0371 4.0283 4.0047 3.9873 3.9645 3.9413 3.9515 3.9824 4.0022 3.9996 3.9690 3.9208 3.8792 3.8483 3.8223 3.8188 3.8511 3.8860 3.8773 3.8441 3.8372 3.8458 3.8453 3.8374 3.8258 3.7941 3.7560 3.7504 3.7584 3.7720 3.8059 3.8312 3.8020 3.7679 3.7848 3.8008 3.7833 3.7682 3.7858 3.8391 3.8853 3.8705 3.8409 3.8636 3.8996 3.8944 3.8579 3.8566 3.9188 4.0062 4.0474 4.0027 3.9218 3.8811 3.9004 3.9594 4.0117 4.0281 4.0135 3.9890 3.9866 4.0231 4.0668 4.0711 4.0718 4.0907 4.1074 4.1136 4.1138 4.1124 4.1224 4.1473 4.1677 4.1812 4.1931 4.2098 4.2496 4.2805 4.2600 4.2349 4.2179 4.1955 4.1807 4.1928 4.2267 4.2628 4.2767 4.2725 4.2779 4.2896 4.3012 4.3192 4.3254 4.3072 4.2831 4.2739 4.2812 4.2903 4.2892 4.2842 4.2814 4.2716 4.2685 4.2888 4.3127 4.3113 4.2756 4.2453 4.2532 4.2708 4.2649 4.2519 4.2465 4.2432 4.2347 4.2358 4.2439 4.2367 4.2059 4.1670 4.1390 4.1169 4.0873 4.0583 4.0420 4.0230 3.9962 3.9904 3.9978 3.9931 3.9969 3.9990 3.9734 3.9391 3.9346 3.9351 3.9215 3.9048 3.8817 3.8531 3.8234 3.7945 3.7757 3.7681 3.7504 3.7561 3.8093 3.8333 3.8044 3.7842 3.7740 3.7390 3.6953 3.7004 3.7884 3.8750 3.8778 3.8252 3.7715 3.7343 3.7136 3.7297 3.7714 3.7938 3.7986 3.8132 3.8214 3.8113 3.8082 3.8193 3.8296 3.8395 3.8442 3.8188 3.8019 3.8431 3.8958 3.9200 3.9416 3.9829 4.0340 4.0593 4.0356 3.9976 3.9609 3.9166 3.8948 3.9234 3.9567 3.9652 3.9733 3.9786 3.9787 4.0049 4.0699 4.1040 4.0790 4.0643 4.0772 4.0804 4.0748 4.0894 4.1372 4.1957 4.2248 4.2129 4.1991 4.2083 4.2033 4.1846 4.1891 4.2261 4.2594 4.2756 4.2879 4.2997 4.2884 4.2581 4.2359 4.2251 4.2259 4.2409 4.2656 4.2852 4.2798 4.2515 4.2234 4.2279 4.2725 4.3213 4.3505 4.3574 4.3353 4.2998 4.2797 4.2735 4.2693 4.2616 4.2305 4.1807 4.1626 4.1935 4.2462 4.2942 4.2951 4.2192 4.1520 4.1463 4.1483 4.1176 4.0722 4.0506 4.0581 4.0586 4.0511 4.0593 4.0653 4.0676 4.0712 4.0433 3.9804 3.9430 3.9524 3.9672 3.9626 3.9507 3.9430 3.9276 3.9017 3.8604 3.8038 3.7741 3.7966 3.8357 3.8557 3.8527 3.8440 3.8220 3.7784 3.7385 3.7200 3.7338 3.7643 3.7726 3.7769 3.7994 3.8130 3.8246 3.8414 3.8317 3.8082 3.7971 3.7695 3.7312 3.7098 3.7156 3.7561 3.7995 3.8393 3.8735 3.8685 3.8533 3.8750 3.9050 3.8990 3.8768 3.8811 3.9113 3.9429 3.9672 3.9982 4.0258 4.0165 3.9839 3.9711 3.9829 4.0049 4.0441 4.0756 4.0873 4.1022 4.1212 4.1225 4.0995 4.0704 4.0589 4.0752 4.0984 4.1119 4.1343 4.1586 4.1726 4.1798 4.1820 4.1855 4.2094 4.2572 4.2993 4.3176 4.3228 4.3257 4.3240 4.3044 4.2854 4.2829 4.2876 4.2851 4.2630 4.2451 4.2550 4.2679 4.2826 4.3160 4.3282 4.2968 4.2558 4.2432 4.2685 4.3072 4.3280 4.3404 4.3346 4.2759 4.1998 4.1702 4.2124 4.2732 4.2847 4.2629 4.2425 4.2098 4.1770 4.1763 4.1911 4.1882 4.1759 4.1663 4.1330 4.0891 4.0682 4.0682 4.0624 4.0448 4.0438 4.0718 4.0979 4.0821 4.0266 3.9864 4.0021 4.0353 4.0288 4.0025 3.9863 3.9888 4.0062 4.0005 3.9590 3.9127 3.8824 3.8720 3.8572 3.8300 3.8234 3.8445 3.8487 3.8179 3.7819 3.7594 3.7596 3.7757 3.7787 3.7721 3.7664 3.7472 3.7242 3.7326 3.7628 3.7598 3.7196 3.6824 3.6889 3.7206 3.7392 3.7456 3.7603 3.7863 3.8155 3.8447 3.8480 3.8398 3.8611 3.8849 3.8789 3.8712 3.8924 3.9250 3.9419 3.9403 3.9318 3.9006 3.8682 3.9011 3.9809 4.0246 4.0353 4.0516 4.0705 4.0590 4.0341 4.0256 4.0151 4.0109 4.0392 4.0860 4.1355 4.1624 4.1581 4.1609 4.1803 4.2018 4.2184 4.2250 4.2147 4.1899 4.1642 4.1482 4.1911 4.2695 4.2826 4.2485 4.2136 4.1887 4.2074 4.2838 4.3433 4.3051 4.1978 4.1125 4.1055 4.1904 4.2885 4.3294 4.3390 4.3447 4.3437 4.3337 4.2939 4.2301 4.1919 4.2039 4.2005 4.1584 4.1648 4.2182 4.2373 4.2478 4.2490 4.2036 4.1512 4.1438 4.1641 4.1804 4.1629 4.1034 4.0591 4.0682 4.0891 4.0944 4.1156 4.1410 4.1247 4.0698 4.0288 4.0200 4.0364 4.0512 4.0237 3.9955 4.0102 4.0145 3.9760 3.9662 4.0031 3.9971 3.9255 3.8581 3.8404 3.8401 3.8177 3.7523 3.7374 3.8234 3.8702 3.8596 3.8727 3.9066 3.8972 3.8594 3.8644 3.9048 3.9325 3.9488 3.9425 3.8726 3.7838 3.7260 3.7145 3.7316 3.7265 3.7039 3.7232 3.7572 3.7269 3.6752 3.6567 3.6789 3.7518 3.8376 3.8883 3.9189 3.9443 3.9508 3.9067 3.8554 3.9047 4.0260 4.0644 4.0403 4.0125 3.9892 4.0571 4.1430 4.0919 3.9942 3.9804 4.0280 4.0635 4.1130 4.2038 4.2403 4.2103 4.1852 4.1776 4.1395 4.0567 4.0227 4.1071 4.1994 4.2576 4.3552 4.4440 4.4127 4.3346 4.3393 4.3957 4.3489 4.1848 4.0975 4.0920 4.0744 4.1138 4.2179 4.2339 4.1554 4.1073 4.1208 4.0668 3.9841 4.0039 4.0003 3.9251 3.9485 4.0722 4.1164 4.0952 4.1228 4.2088 4.2388 4.2068 4.2166 4.2125 4.1202 4.0571 4.0532 4.1558 4.3152 4.2978 4.1616 4.1443 4.1086 4.0384 4.0687 4.0723 3.9987 3.9775 3.9429 3.9011 3.8918 3.8856 3.8072 3.6220 3.5458 3.7027 3.8685 3.8723 3.8022 3.7797 3.6861 3.4658 3.3700 3.4747 3.5662 3.6115 3.6012 3.4953 3.5476 3.7015 3.6951 3.5577 3.4871 3.4478 3.3931 3.3936 3.3991 3.3322 3.1924 3.0838 3.1296 3.3243 3.5037 3.5536 3.5758 3.5586 3.4281 3.3716 3.4918 3.5988 3.6349 3.5910 3.5086 3.4555 3.4986 3.6916 3.8286 3.7717 3.6641 3.6092 3.6049 3.6598 3.7989 3.8886 3.6724 3.4683 3.5635 3.7365 3.8570 3.9400 3.9338 3.8813 3.9444 4.1208 4.2367 4.1597 4.0182 3.9569 3.9355 3.9143 3.8876 3.9056 3.9516 3.9832 4.0654 4.1691 4.2439 4.2878 4.2749 4.2742 4.3295 4.3168 4.1415 3.9372 3.9371 4.1325 4.3543 4.4586 4.4060 4.3445 4.3805 4.3726 4.2661 4.1896 4.1046 3.9667 3.8929 4.0597 4.3152 4.3867 4.3182 4.2544 4.2140 4.1667 4.1149 4.0481 4.0083 4.0227 4.0487 4.1225 4.1894 4.1586 4.1040 4.0811 4.0557 4.1132 4.2713 4.3924 4.3247 4.1275 4.0199 4.0101 4.0239 4.0590 4.0638 3.9718 3.8246 3.7107 3.7541 3.9144 3.9891 3.9794 3.8180 3.6486 3.6818 3.7822 3.8763 4.0135 4.1085 4.0391 3.9151 3.8199 3.6551 3.4665 3.4429 3.5590 3.6877 3.7650 3.7755 3.7277 3.6902 3.6983 3.7011 3.6887 3.7134 3.7730 3.7947 3.7823 3.7317 3.6262 3.5784 3.5698 3.4989 3.5104 3.6605 3.7955 3.9019 3.9897 4.0021 3.9406 3.8701 3.7931 3.6613 3.5998 3.6305 3.6473 3.7123 3.8283 3.9245 4.0329 4.0829 4.0211 4.0096 4.1040 4.1869 4.2066 4.1136 3.9039 3.7299 3.7298 3.8234 3.8948 3.9680 4.0104 3.9448 3.8663 3.8502 3.8368 3.8798 4.0532 4.2471 4.3475 4.2682 4.1016 4.1576 4.4150 4.5706 4.5509 4.4277 4.3082 4.2466 4.1368 4.0491 4.0839 4.1753 4.2691 4.2812 4.1577 4.0094 3.9138 3.9199 4.0023 4.0903 4.2195 4.3667 4.4259 4.4122 4.4122 4.3466 4.1514 3.9618 3.9475 4.0487 4.1192 4.1147 4.0907 4.1024 4.1024 4.0307 3.9274 3.8137 3.7349 3.7556 3.8267 3.8344 3.8090 3.8208 3.9034 4.0498 4.1513 4.1832 4.2183 4.1704 3.9715 3.7718 3.6723 3.5610 3.4824 3.5552 3.6635 3.6271 3.5437 3.6304 3.7303 3.7012 3.6188 3.5569 3.5710 3.6532 3.7184 3.7485 3.7741 3.7638 3.7374 3.7146 3.6672 3.6052 3.6175 3.7730 3.8550 3.6726 3.4110 3.4656 3.8464 4.1414 4.1006 3.9079 3.8500 3.9691 4.0257 3.8513 3.6397 3.6523 3.8355 3.9043 3.8691 3.9021 3.9751 4.0474 4.0936 4.1213 4.1933 4.3028 4.4050 4.3897 4.2845 4.2412 4.1160 3.9072 3.9179 4.1168 4.3097 4.3819 4.3499 4.2688 4.1972 4.1233 4.0957 4.1848 4.2591 4.2379 4.2120 4.1625 4.0897 4.0789 4.1586 4.2863 4.3210 4.2028 4.0356 3.8960 3.8457 3.9294 4.0042 3.8854 3.7650 3.7648 3.7415 3.5443 3.3887 3.3969 3.3726 3.2292 3.0580 2.9826 2.9791 2.9773 2.9272 2.7399 2.5032 2.3386 2.2621 2.1826 2.0198 1.7527 1.3994 1.0421 0.8313 0.8379 1.0071 1.1620 1.1241 0.8451 0.5447 0.3268 0.0256 -0.3426 -0.5481 -0.5391 -0.4992 -0.6288 -0.8772 -1.0927 -1.3706 -1.9960 -2.6370 -2.8954 -3.0923 -3.2882 -3.3825 -3.5081 -3.5893 -3.6076 -3.7203 -3.8884 -3.9815 -4.0522 -4.0332 -3.9273 -4.2025 -4.8333 -5.4544 -5.7822 -5.4846 -4.8133 -4.6546 -5.0683 -5.6868 -5.9887 -5.6421 -5.0674 -4.9115 -4.9862 -5.0753 -5.4126 -5.7638 -5.3747 -4.3985 -3.9502 -4.3882 -4.6642 -4.0590 -3.2857 -3.0553 -3.0171 -2.9334 -2.6387 -2.3332 -2.4326 -2.7259 -2.8058 -2.7649 -2.7696 -2.8219 -2.9168 -3.0049 -2.9978 -2.8766 -2.7459 -2.7877 -2.7952 -2.6307 -2.4433 -2.6332 -3.1869 -3.2425 -2.2819 -1.0299 -0.2065 -0.0031 -0.2092 -0.4339 -0.9713 -1.9048 -2.3555 -2.1326 -2.0028 -2.2714 -2.4559 -2.1483 -1.5499 -1.1712 -1.4571 -2.2413 -2.4907 -2.1361 -1.7915 -1.7738 -1.9718 -2.1032 -1.8157 -1.2609 -1.0888 -1.4205 -1.6319 -1.4174 -1.0870 -0.6456 0.0054 0.4237 0.3262 0.0030 -0.1614 0.0100 0.1762 0.0037 -0.4210 -0.8006 -1.0810 -1.2003 -1.0901 -0.9077 -0.6167 -0.0365 0.6056 0.6823 0.2631 0.0315 0.1720 0.5950 0.9670 1.2035 1.0868 0.7313 0.2079 -0.4428 -1.0584 -1.3108 -0.8113 -0.1252 -0.0700 -0.4471 -0.4977 -0.0009 0.4090 0.4309 0.3372 0.3134 0.3018 0.2507 -0.0140 -0.2245 -0.2516 -0.0333 0.2052 0.3167 0.3388 0.2979 0.2408 0.2438 0.1288 0.0355 -0.0726 0.1884 0.5430 0.5037 0.3928 0.6536 0.9805 1.0402 0.8571 0.8218 0.8197 0.7281 0.5522 0.6925 1.0045 1.0707 0.6734 0.6085 1.0526 1.2845 1.1089 1.2953 1.3806 1.1912 0.9422 0.8630 0.9159 1.4254 1.8088 1.7759 1.2884 0.9880 0.8059 0.6631 0.3597 0.2877 0.5101 0.9324 1.1062 1.1189 1.0873 1.2963 1.4214 1.6282 1.7053 1.4973 0.9790 1.0983 1.3852 1.5429 1.5234 1.5292 1.3104 1.2757 0.8869 0.5145 0.7113 1.3543 1.3890 0.9667 0.7196 1.4537 2.0366 2.1264 1.8825 1.6793 1.3166 1.3493 1.3008 1.2872 1.2882 1.6478 1.8070 2.1904 2.2762 2.0932 1.6617 1.7195 2.0066 2.6213 2.6506 2.5683 2.3499 2.5807 2.5199 2.4837 1.9532 1.7694 1.8284 2.2359 2.1669 2.0885 1.9388 2.2269 2.5849 3.0879 2.8728 3.2147 3.0896 2.6566 1.8906 1.8568 1.8620 1.9197 0.5811 0.0707 -0.1140 -0.2507 -0.3872 -0.5261 -0.3324 0.6124 0.6609 0.7726 0.7911 0.8297 0.4739 0.2283 0.2148 0.4259 0.5007 -------------------------------  From telecom@delta.eecs.nwu.edu Fri Sep 3 23:17:45 1993 Received: from delta.eecs.nwu.edu by gaak.LCS.MIT.EDU via TCP with SMTP id AA17207; Fri, 3 Sep 93 23:17:40 EDT Received: by delta.eecs.nwu.edu id AA00985 (5.65c/IDA-1.4.4 for ptownson@gaak.lcs.mit.edu); Fri, 3 Sep 1993 22:17:34 -0500 Date: Fri, 3 Sep 1993 22:17:34 -0500 From: TELECOM Moderator Message-Id: <199309040317.AA00985@delta.eecs.nwu.edu> To: ptownson@gaak.LCS.MIT.EDU Subject: AT&T TrueVoice DSP Analysis Status: O >From telecom Fri Sep 3 22:15:46 1993 Received: by delta.eecs.nwu.edu id AA10360 (5.65c/IDA-1.4.4 for telecom); Fri, 3 Sep 1993 22:15:46 -0500 Date: Fri, 3 Sep 1993 22:15:46 -0500 From: TELECOM Moderator Message-Id: <199309040315.AA10360@delta.eecs.nwu.edu> To: telecom Subject: AT&T TrueVoice DSP Analysis Status: R A special report for the readers, submitted by malcolm@apple.com. I hope you find it useful and interesting. PAT Subject: AT&T TrueVoice DSP Analysis Date: Fri, 3 Sep 93 13:39:17 -0700 From: malcolm@apple.com I've just finished an analysis of the AT&T TrueVoice demo that has been advertised by AT&T and discussed in the TELECOM Digest. Here are the results of the analysis. First, there is a reason that the TrueVoice version sounds louder. Above 300Hz there is a pretty uniform gain of 4dB. Between 100-300 Hz there is a 14 dB gain peak (with a notch at 180 Hz to kill the third harmonic of the power line hum.) Here are the details: 0-100 Hz Not much change. 100-300Hz A broad gain bump of about 14dB 180Hz Deep 12.5dB notch (back to 1.5 dB) 300-4kHz 4dB extra gain with a .5dB ripple Method: The AT&T TrueVoice demo (1 800 932-2000) was digitized with the Telecom pod on the Macintosh Quadra 840AV (using the AT&T 3210 DSP chip). Samples were converted with 16 bits of linear resolution, sampled at 8kHz, and stored in an AIFF file. All data was collected from a single phone call using an analog phone line. The results were loaded into Matlab on a Macintosh for analysis. The TrueVoice demo starts with 30 seconds of normal speech and then automatically switches on the TrueVoice modification and replays the same speech. The beginning of each demo sample was aligned by eye, and then a cross correlation was used to refine the relative starting positions. All further analysis was done on these aligned samples of modified and unmodified speech. Note, the TrueVoice modification is turned on over a second or two. (I don't know why AT&T does this. Perhaps they didn't want the change to seem too jarring.) The cross correlation analysis was done using these first couple of seconds: this should make the correlation more accurate. The spectral analysis that followed skipped the first three seconds. A periodogram analysis was performed using a window of data of length 2048 and skipping 16 samples between windows. A Hamming window was used before FFT'ing and the power spectral density was computed. The power spectral density for all windows was summed (averaged) to arrive at a robust estimate of the power spectral density of the entire sample. A total of 170,000 data samples was used from each example. The power spectral density for each example (unmodified and modified speech) was converted to dB. The two PSDs are then subtracted to estimate the difference in spectral energy due to the TrueVoice filtering. Results: The broad filter characteristics are shown below. 0-100 Hz Not much change. 100-300Hz A broad gain of 13.7 dB, 3dB Q is 1.5 120Hz About a 1dB dip in the gain 180Hz Narrow band reject down to 1.5 dB, 3dB Q is 18 300-4kHz 4dB extra gain with a .5dB ripple Note, there is a deep notch at the 3rd harmonic of the power line frequency and a extra small dip at the second harmonic. I don't have enough information to tell if they are doing anything different at 60Hz. The filtering appears to be stationary. The demo over the phone turns the filtering on gradually, but after a second there appears to be no change in the filter's characteristics. The analysis described above was performed over several subsections of the data and there was no visible change in the filter's characteristics. Acknowledgements: Many thanks to Stuart Davidson for setting up the hardware and software to digitize the demo. Thanks also to Michele Covell (Interval Research) for help with the filter analysis. The periodogram is discussed in Larry Marple's book "Digital Spectral Analysis with Applications." Disclaimer: I have no affiliation with AT&T except as a mostly satisfied customer. This work is not an official Apple project. TrueVoice and AT&T are undoubtably trademarks of AT&T. Malcolm Slaney malcolm@apple.com =================== Matlab Periodogram Code ============================ % function y=periodogram(x,width,skip) % Compute the periodogram (an estimate of the power spectral % density) of an array. The width of each analysis defaults to % 256 points, and the default skip between windows is half the % window size. function y=periodogram(x,width,skip) if (nargin < 2) width = 256; end if (nargin < 3) skip = width/2; end nwinds = fix((length(x)-width)/skip); x = reshape(x,length(x),1); wind = hamming(width); y = zeros(width,1); for i=1:nwinds firstPoint = (i-1)*skip + 1; d = [x(firstPoint:firstPoint+width-1).*wind;zeros(width,1)]; f = fft(d); f = f(1:width); y = y + f .* conj(f); end y = y/nwinds; =============== Matlab TrueVoice Analysis Code ========================= % Analyze a recorded sample of the AT&T TrueVoice demo. % This script assumes that the entire demo is stored in % an array called tv. The start of the first and second % version of the demo are indicated by the first and second % pointers. (These will be different for anybody else's % recording of the phone call.) first=116000; second=395500+41; firstEnd = 355000; secondEnd = 610000; sampleRate = 8000; %segment = 170000; segment = 32768; offset = 30000; if (second + offset + segment > secondEnd) disp('Segment is past end of second example.'); return; end s1=tv(first+offset:first+offset+segment-1); s2=tv(second+offset:second+offset+segment-1); % Optionally compute the optimum cross correlation. % If the alignment is done right, then the maximum % value of the cross correlation should be in the % middle of the cross correlation or sample number % equal to segment size . % cor = xcorr(s1,s2); % plot(cor) % [m,i] = max(cor); % i r1 = 10*log10(periodogram(s1, 512, 128)); r2 = 10*log10(periodogram(s2, 512, 128)); f = (0:length(r1)-1)/length(r1)*sampleRate/2; =============== Spectral Analysis of TrueVoice ========================= % Spectral analysis (in dB) of the AT&T TrueVoice demo. % The results are shown below at 1.9531Hz intervals from % 0Hz to nearly 4kHz. -0.0156 -0.0156 -0.0154 -0.0138 2.8762 1.5637 1.3204 0.2571 0.6921 0.9584 0.7187 -0.7438 -1.7550 -2.0471 -2.0754 -1.9768 -1.6302 -0.7998 0.7231 0.9736 0.9899 0.9882 0.9827 0.9418 0.2723 -0.3172 -0.4245 0.0216 -0.1162 -0.1018 -0.0875 -0.0747 -0.0601 -0.0360 0.1168 0.9137 2.4710 2.5066 3.5501 2.9577 3.3023 3.0256 3.8413 4.5084 6.3061 6.4052 7.6116 7.6745 8.5071 8.4967 9.0233 9.0818 9.5900 9.7797 10.1627 10.3327 10.6086 10.7367 10.9011 10.8782 10.5145 10.0140 10.3593 11.2615 11.8157 11.9453 12.0295 12.0388 12.0603 12.1355 12.3434 12.4985 12.6426 12.7032 12.8563 12.9541 13.0520 13.0989 13.2080 13.2716 13.3565 13.3807 13.4805 13.5283 13.5936 13.6157 13.6703 13.6777 13.6661 12.1029 5.5750 2.3909 1.6613 2.2414 4.9604 11.2490 13.7420 13.7310 13.7230 13.6977 13.6595 13.5861 13.5226 13.4833 13.4672 13.4355 13.3998 13.3534 13.3033 13.2455 13.1833 13.1151 13.0436 12.9571 12.8699 12.7882 12.6880 12.5852 12.5082 12.4377 12.3462 12.2276 12.1066 11.9765 11.8630 11.7580 11.6211 11.4772 11.3559 11.2418 11.0955 10.9580 10.8388 10.6905 10.5217 10.3533 10.2052 10.0873 9.9788 9.8533 9.7000 9.5401 9.3863 9.2267 9.0551 8.8746 8.7166 8.5856 8.4393 8.2574 8.0735 7.8216 7.2432 6.4982 6.2854 6.7300 7.0660 7.0194 6.8627 6.6837 6.4944 6.3324 6.1924 6.0580 5.9400 5.8304 5.6939 5.5120 5.3422 5.2091 5.0764 4.9489 4.8327 4.7075 4.5998 4.4920 4.3769 4.2938 4.2189 4.1290 4.0337 3.9499 3.8720 3.7967 3.7345 3.6756 3.6097 3.5443 3.4910 3.4470 3.4026 3.3690 3.3538 3.3306 3.2824 3.2459 3.2375 3.2455 3.2562 3.2654 3.2769 3.2867 3.2941 3.3026 3.3110 3.3201 3.3315 3.3440 3.3697 3.4085 3.4461 3.4825 3.5228 3.5549 3.5623 3.5717 3.6202 3.6865 3.7361 3.7728 3.7982 3.8225 3.8617 3.8934 3.9120 3.9411 3.9805 4.0103 4.0313 4.0423 4.0499 4.0719 4.1125 4.1634 4.1984 4.2129 4.2306 4.2533 4.2706 4.2897 4.3161 4.3271 4.3166 4.3097 4.3212 4.3222 4.3041 4.3011 4.3121 4.3238 4.3343 4.3339 4.3267 4.3170 4.2965 4.2704 4.2473 4.2342 4.2294 4.2322 4.2381 4.2338 4.2211 4.2096 4.2020 4.1905 4.1708 4.1571 4.1513 4.1370 4.1146 4.0957 4.0799 4.0686 4.0589 4.0159 3.9430 3.9157 3.9485 3.9754 3.9758 3.9720 3.9625 3.9376 3.9162 3.9101 3.9068 3.9034 3.9000 3.8920 3.8796 3.8663 3.8585 3.8588 3.8562 3.8374 3.8205 3.8194 3.8240 3.8239 3.8113 3.8002 3.8025 3.8091 3.8143 3.8212 3.8278 3.8367 3.8441 3.8418 3.8338 3.8264 3.8311 3.8473 3.8638 3.8787 3.8898 3.8965 3.8994 3.8999 3.9121 3.9445 3.9715 3.9803 3.9848 3.9975 4.0300 4.0597 4.0501 4.0212 4.0095 4.0264 4.0668 4.1108 4.1344 4.1324 4.1079 4.0761 4.0591 4.0615 4.0805 4.1049 4.1293 4.1566 4.1745 4.1746 4.1766 4.1942 4.2163 4.2459 4.2697 4.2636 4.2516 4.2672 4.2902 4.2960 4.2897 4.2854 4.2939 4.3081 4.3115 4.3180 4.3467 4.3717 4.3674 4.3452 4.3170 4.2937 4.2886 4.2898 4.2866 4.2869 4.2744 4.2453 4.2252 4.2041 4.1907 4.1994 4.2001 4.1973 4.2036 4.2054 4.1940 4.1572 4.1137 4.1197 4.1507 4.1396 4.1036 4.0570 4.0095 3.9778 3.9774 3.9987 4.0189 4.0231 4.0033 3.9736 3.9444 3.9261 3.9289 3.9371 3.9147 3.8360 3.7424 3.7082 3.7061 3.6903 3.6718 3.6725 3.6802 3.6756 3.6680 3.6843 3.7143 3.7304 3.7240 3.7022 3.6766 3.6636 3.6838 3.7364 3.7735 3.7715 3.7745 3.8092 3.8338 3.8216 3.8117 3.8196 3.8251 3.8316 3.8483 3.8663 3.8788 3.8834 3.8788 3.8763 3.8858 3.8962 3.8959 3.8940 3.9102 3.9491 3.9801 3.9850 4.0079 4.0531 4.0789 4.0709 4.0647 4.0794 4.0913 4.0958 4.1114 4.1341 4.1524 4.1771 4.1970 4.2059 4.2127 4.2154 4.2141 4.2150 4.2213 4.2226 4.2111 4.1941 4.1933 4.2252 4.2576 4.2572 4.2150 4.1531 4.1139 4.1130 4.1501 4.1874 4.1887 4.1605 4.1311 4.1236 4.1473 4.1852 4.2052 4.2233 4.2426 4.2489 4.2538 4.2531 4.2337 4.2260 4.2384 4.2403 4.2367 4.2453 4.2668 4.2789 4.2627 4.2359 4.2013 4.1635 4.1479 4.1810 4.2237 4.2029 4.1435 4.1030 4.0890 4.0912 4.1109 4.1343 4.1405 4.1113 4.0482 3.9993 3.9791 3.9588 3.9272 3.8873 3.8515 3.8288 3.8098 3.8065 3.8195 3.8113 3.7923 3.8059 3.8273 3.8364 3.8258 3.7813 3.7552 3.7867 3.8330 3.8302 3.8012 3.7939 3.7789 3.7384 3.7168 3.7019 3.6732 3.6598 3.6861 3.7230 3.7378 3.7430 3.7643 3.7819 3.7818 3.7938 3.7803 3.7341 3.7190 3.7303 3.7457 3.7783 3.8146 3.8645 3.9324 3.9773 4.0000 4.0181 3.9884 3.9119 3.8909 3.9387 4.0071 4.0808 4.1228 4.1229 4.1029 4.0704 4.0309 4.0110 4.0390 4.0767 4.0965 4.1365 4.1915 4.2151 4.2345 4.2727 4.2879 4.2576 4.2202 4.2258 4.2612 4.2724 4.2555 4.2427 4.2471 4.2661 4.2896 4.3010 4.3005 4.2957 4.2855 4.2740 4.2666 4.2695 4.2819 4.2873 4.2675 4.2188 4.1749 4.1607 4.1578 4.1489 4.1283 4.1087 4.1226 4.1646 4.1975 4.2025 4.1791 4.1383 4.1318 4.1720 4.1981 4.1785 4.1589 4.1588 4.1699 4.1828 4.1738 4.1261 4.0842 4.0948 4.1499 4.1747 4.1536 4.1222 4.0861 4.0489 4.0195 4.0093 4.0231 4.0133 3.9700 3.9524 3.9736 4.0181 4.0664 4.1272 4.1630 4.1220 4.0943 4.0931 4.0471 3.9736 3.9507 3.9191 3.8515 3.8284 3.8700 3.9406 3.9809 3.9753 3.9601 3.9425 3.9219 3.9120 3.8903 3.8508 3.8170 3.7936 3.7934 3.8167 3.8262 3.8181 3.8137 3.7817 3.7410 3.7353 3.7615 3.8004 3.8142 3.8099 3.8200 3.8412 3.8395 3.8060 3.7680 3.7490 3.7546 3.7850 3.8190 3.8374 3.8329 3.8101 3.7999 3.8239 3.8626 3.8995 3.9293 3.9370 3.9439 3.9866 4.0243 4.0340 4.0458 4.0547 4.0425 4.0240 4.0253 4.0536 4.0953 4.1350 4.1705 4.1965 4.2225 4.2664 4.2941 4.2819 4.2690 4.2711 4.2839 4.2996 4.3138 4.3118 4.2843 4.2663 4.2734 4.2854 4.2986 4.3212 4.3424 4.3508 4.3505 4.3336 4.3145 4.3226 4.3632 4.4118 4.4020 4.3286 4.2562 4.2137 4.2224 4.2605 4.2589 4.2231 4.2219 4.2602 4.2963 4.2905 4.2564 4.2230 4.2089 4.2149 4.2181 4.2004 4.1570 4.0962 4.0639 4.0803 4.0981 4.1046 4.0952 4.0567 4.0047 3.9813 3.9860 3.9971 4.0172 4.0371 4.0283 4.0047 3.9873 3.9645 3.9413 3.9515 3.9824 4.0022 3.9996 3.9690 3.9208 3.8792 3.8483 3.8223 3.8188 3.8511 3.8860 3.8773 3.8441 3.8372 3.8458 3.8453 3.8374 3.8258 3.7941 3.7560 3.7504 3.7584 3.7720 3.8059 3.8312 3.8020 3.7679 3.7848 3.8008 3.7833 3.7682 3.7858 3.8391 3.8853 3.8705 3.8409 3.8636 3.8996 3.8944 3.8579 3.8566 3.9188 4.0062 4.0474 4.0027 3.9218 3.8811 3.9004 3.9594 4.0117 4.0281 4.0135 3.9890 3.9866 4.0231 4.0668 4.0711 4.0718 4.0907 4.1074 4.1136 4.1138 4.1124 4.1224 4.1473 4.1677 4.1812 4.1931 4.2098 4.2496 4.2805 4.2600 4.2349 4.2179 4.1955 4.1807 4.1928 4.2267 4.2628 4.2767 4.2725 4.2779 4.2896 4.3012 4.3192 4.3254 4.3072 4.2831 4.2739 4.2812 4.2903 4.2892 4.2842 4.2814 4.2716 4.2685 4.2888 4.3127 4.3113 4.2756 4.2453 4.2532 4.2708 4.2649 4.2519 4.2465 4.2432 4.2347 4.2358 4.2439 4.2367 4.2059 4.1670 4.1390 4.1169 4.0873 4.0583 4.0420 4.0230 3.9962 3.9904 3.9978 3.9931 3.9969 3.9990 3.9734 3.9391 3.9346 3.9351 3.9215 3.9048 3.8817 3.8531 3.8234 3.7945 3.7757 3.7681 3.7504 3.7561 3.8093 3.8333 3.8044 3.7842 3.7740 3.7390 3.6953 3.7004 3.7884 3.8750 3.8778 3.8252 3.7715 3.7343 3.7136 3.7297 3.7714 3.7938 3.7986 3.8132 3.8214 3.8113 3.8082 3.8193 3.8296 3.8395 3.8442 3.8188 3.8019 3.8431 3.8958 3.9200 3.9416 3.9829 4.0340 4.0593 4.0356 3.9976 3.9609 3.9166 3.8948 3.9234 3.9567 3.9652 3.9733 3.9786 3.9787 4.0049 4.0699 4.1040 4.0790 4.0643 4.0772 4.0804 4.0748 4.0894 4.1372 4.1957 4.2248 4.2129 4.1991 4.2083 4.2033 4.1846 4.1891 4.2261 4.2594 4.2756 4.2879 4.2997 4.2884 4.2581 4.2359 4.2251 4.2259 4.2409 4.2656 4.2852 4.2798 4.2515 4.2234 4.2279 4.2725 4.3213 4.3505 4.3574 4.3353 4.2998 4.2797 4.2735 4.2693 4.2616 4.2305 4.1807 4.1626 4.1935 4.2462 4.2942 4.2951 4.2192 4.1520 4.1463 4.1483 4.1176 4.0722 4.0506 4.0581 4.0586 4.0511 4.0593 4.0653 4.0676 4.0712 4.0433 3.9804 3.9430 3.9524 3.9672 3.9626 3.9507 3.9430 3.9276 3.9017 3.8604 3.8038 3.7741 3.7966 3.8357 3.8557 3.8527 3.8440 3.8220 3.7784 3.7385 3.7200 3.7338 3.7643 3.7726 3.7769 3.7994 3.8130 3.8246 3.8414 3.8317 3.8082 3.7971 3.7695 3.7312 3.7098 3.7156 3.7561 3.7995 3.8393 3.8735 3.8685 3.8533 3.8750 3.9050 3.8990 3.8768 3.8811 3.9113 3.9429 3.9672 3.9982 4.0258 4.0165 3.9839 3.9711 3.9829 4.0049 4.0441 4.0756 4.0873 4.1022 4.1212 4.1225 4.0995 4.0704 4.0589 4.0752 4.0984 4.1119 4.1343 4.1586 4.1726 4.1798 4.1820 4.1855 4.2094 4.2572 4.2993 4.3176 4.3228 4.3257 4.3240 4.3044 4.2854 4.2829 4.2876 4.2851 4.2630 4.2451 4.2550 4.2679 4.2826 4.3160 4.3282 4.2968 4.2558 4.2432 4.2685 4.3072 4.3280 4.3404 4.3346 4.2759 4.1998 4.1702 4.2124 4.2732 4.2847 4.2629 4.2425 4.2098 4.1770 4.1763 4.1911 4.1882 4.1759 4.1663 4.1330 4.0891 4.0682 4.0682 4.0624 4.0448 4.0438 4.0718 4.0979 4.0821 4.0266 3.9864 4.0021 4.0353 4.0288 4.0025 3.9863 3.9888 4.0062 4.0005 3.9590 3.9127 3.8824 3.8720 3.8572 3.8300 3.8234 3.8445 3.8487 3.8179 3.7819 3.7594 3.7596 3.7757 3.7787 3.7721 3.7664 3.7472 3.7242 3.7326 3.7628 3.7598 3.7196 3.6824 3.6889 3.7206 3.7392 3.7456 3.7603 3.7863 3.8155 3.8447 3.8480 3.8398 3.8611 3.8849 3.8789 3.8712 3.8924 3.9250 3.9419 3.9403 3.9318 3.9006 3.8682 3.9011 3.9809 4.0246 4.0353 4.0516 4.0705 4.0590 4.0341 4.0256 4.0151 4.0109 4.0392 4.0860 4.1355 4.1624 4.1581 4.1609 4.1803 4.2018 4.2184 4.2250 4.2147 4.1899 4.1642 4.1482 4.1911 4.2695 4.2826 4.2485 4.2136 4.1887 4.2074 4.2838 4.3433 4.3051 4.1978 4.1125 4.1055 4.1904 4.2885 4.3294 4.3390 4.3447 4.3437 4.3337 4.2939 4.2301 4.1919 4.2039 4.2005 4.1584 4.1648 4.2182 4.2373 4.2478 4.2490 4.2036 4.1512 4.1438 4.1641 4.1804 4.1629 4.1034 4.0591 4.0682 4.0891 4.0944 4.1156 4.1410 4.1247 4.0698 4.0288 4.0200 4.0364 4.0512 4.0237 3.9955 4.0102 4.0145 3.9760 3.9662 4.0031 3.9971 3.9255 3.8581 3.8404 3.8401 3.8177 3.7523 3.7374 3.8234 3.8702 3.8596 3.8727 3.9066 3.8972 3.8594 3.8644 3.9048 3.9325 3.9488 3.9425 3.8726 3.7838 3.7260 3.7145 3.7316 3.7265 3.7039 3.7232 3.7572 3.7269 3.6752 3.6567 3.6789 3.7518 3.8376 3.8883 3.9189 3.9443 3.9508 3.9067 3.8554 3.9047 4.0260 4.0644 4.0403 4.0125 3.9892 4.0571 4.1430 4.0919 3.9942 3.9804 4.0280 4.0635 4.1130 4.2038 4.2403 4.2103 4.1852 4.1776 4.1395 4.0567 4.0227 4.1071 4.1994 4.2576 4.3552 4.4440 4.4127 4.3346 4.3393 4.3957 4.3489 4.1848 4.0975 4.0920 4.0744 4.1138 4.2179 4.2339 4.1554 4.1073 4.1208 4.0668 3.9841 4.0039 4.0003 3.9251 3.9485 4.0722 4.1164 4.0952 4.1228 4.2088 4.2388 4.2068 4.2166 4.2125 4.1202 4.0571 4.0532 4.1558 4.3152 4.2978 4.1616 4.1443 4.1086 4.0384 4.0687 4.0723 3.9987 3.9775 3.9429 3.9011 3.8918 3.8856 3.8072 3.6220 3.5458 3.7027 3.8685 3.8723 3.8022 3.7797 3.6861 3.4658 3.3700 3.4747 3.5662 3.6115 3.6012 3.4953 3.5476 3.7015 3.6951 3.5577 3.4871 3.4478 3.3931 3.3936 3.3991 3.3322 3.1924 3.0838 3.1296 3.3243 3.5037 3.5536 3.5758 3.5586 3.4281 3.3716 3.4918 3.5988 3.6349 3.5910 3.5086 3.4555 3.4986 3.6916 3.8286 3.7717 3.6641 3.6092 3.6049 3.6598 3.7989 3.8886 3.6724 3.4683 3.5635 3.7365 3.8570 3.9400 3.9338 3.8813 3.9444 4.1208 4.2367 4.1597 4.0182 3.9569 3.9355 3.9143 3.8876 3.9056 3.9516 3.9832 4.0654 4.1691 4.2439 4.2878 4.2749 4.2742 4.3295 4.3168 4.1415 3.9372 3.9371 4.1325 4.3543 4.4586 4.4060 4.3445 4.3805 4.3726 4.2661 4.1896 4.1046 3.9667 3.8929 4.0597 4.3152 4.3867 4.3182 4.2544 4.2140 4.1667 4.1149 4.0481 4.0083 4.0227 4.0487 4.1225 4.1894 4.1586 4.1040 4.0811 4.0557 4.1132 4.2713 4.3924 4.3247 4.1275 4.0199 4.0101 4.0239 4.0590 4.0638 3.9718 3.8246 3.7107 3.7541 3.9144 3.9891 3.9794 3.8180 3.6486 3.6818 3.7822 3.8763 4.0135 4.1085 4.0391 3.9151 3.8199 3.6551 3.4665 3.4429 3.5590 3.6877 3.7650 3.7755 3.7277 3.6902 3.6983 3.7011 3.6887 3.7134 3.7730 3.7947 3.7823 3.7317 3.6262 3.5784 3.5698 3.4989 3.5104 3.6605 3.7955 3.9019 3.9897 4.0021 3.9406 3.8701 3.7931 3.6613 3.5998 3.6305 3.6473 3.7123 3.8283 3.9245 4.0329 4.0829 4.0211 4.0096 4.1040 4.1869 4.2066 4.1136 3.9039 3.7299 3.7298 3.8234 3.8948 3.9680 4.0104 3.9448 3.8663 3.8502 3.8368 3.8798 4.0532 4.2471 4.3475 4.2682 4.1016 4.1576 4.4150 4.5706 4.5509 4.4277 4.3082 4.2466 4.1368 4.0491 4.0839 4.1753 4.2691 4.2812 4.1577 4.0094 3.9138 3.9199 4.0023 4.0903 4.2195 4.3667 4.4259 4.4122 4.4122 4.3466 4.1514 3.9618 3.9475 4.0487 4.1192 4.1147 4.0907 4.1024 4.1024 4.0307 3.9274 3.8137 3.7349 3.7556 3.8267 3.8344 3.8090 3.8208 3.9034 4.0498 4.1513 4.1832 4.2183 4.1704 3.9715 3.7718 3.6723 3.5610 3.4824 3.5552 3.6635 3.6271 3.5437 3.6304 3.7303 3.7012 3.6188 3.5569 3.5710 3.6532 3.7184 3.7485 3.7741 3.7638 3.7374 3.7146 3.6672 3.6052 3.6175 3.7730 3.8550 3.6726 3.4110 3.4656 3.8464 4.1414 4.1006 3.9079 3.8500 3.9691 4.0257 3.8513 3.6397 3.6523 3.8355 3.9043 3.8691 3.9021 3.9751 4.0474 4.0936 4.1213 4.1933 4.3028 4.4050 4.3897 4.2845 4.2412 4.1160 3.9072 3.9179 4.1168 4.3097 4.3819 4.3499 4.2688 4.1972 4.1233 4.0957 4.1848 4.2591 4.2379 4.2120 4.1625 4.0897 4.0789 4.1586 4.2863 4.3210 4.2028 4.0356 3.8960 3.8457 3.9294 4.0042 3.8854 3.7650 3.7648 3.7415 3.5443 3.3887 3.3969 3.3726 3.2292 3.0580 2.9826 2.9791 2.9773 2.9272 2.7399 2.5032 2.3386 2.2621 2.1826 2.0198 1.7527 1.3994 1.0421 0.8313 0.8379 1.0071 1.1620 1.1241 0.8451 0.5447 0.3268 0.0256 -0.3426 -0.5481 -0.5391 -0.4992 -0.6288 -0.8772 -1.0927 -1.3706 -1.9960 -2.6370 -2.8954 -3.0923 -3.2882 -3.3825 -3.5081 -3.5893 -3.6076 -3.7203 -3.8884 -3.9815 -4.0522 -4.0332 -3.9273 -4.2025 -4.8333 -5.4544 -5.7822 -5.4846 -4.8133 -4.6546 -5.0683 -5.6868 -5.9887 -5.6421 -5.0674 -4.9115 -4.9862 -5.0753 -5.4126 -5.7638 -5.3747 -4.3985 -3.9502 -4.3882 -4.6642 -4.0590 -3.2857 -3.0553 -3.0171 -2.9334 -2.6387 -2.3332 -2.4326 -2.7259 -2.8058 -2.7649 -2.7696 -2.8219 -2.9168 -3.0049 -2.9978 -2.8766 -2.7459 -2.7877 -2.7952 -2.6307 -2.4433 -2.6332 -3.1869 -3.2425 -2.2819 -1.0299 -0.2065 -0.0031 -0.2092 -0.4339 -0.9713 -1.9048 -2.3555 -2.1326 -2.0028 -2.2714 -2.4559 -2.1483 -1.5499 -1.1712 -1.4571 -2.2413 -2.4907 -2.1361 -1.7915 -1.7738 -1.9718 -2.1032 -1.8157 -1.2609 -1.0888 -1.4205 -1.6319 -1.4174 -1.0870 -0.6456 0.0054 0.4237 0.3262 0.0030 -0.1614 0.0100 0.1762 0.0037 -0.4210 -0.8006 -1.0810 -1.2003 -1.0901 -0.9077 -0.6167 -0.0365 0.6056 0.6823 0.2631 0.0315 0.1720 0.5950 0.9670 1.2035 1.0868 0.7313 0.2079 -0.4428 -1.0584 -1.3108 -0.8113 -0.1252 -0.0700 -0.4471 -0.4977 -0.0009 0.4090 0.4309 0.3372 0.3134 0.3018 0.2507 -0.0140 -0.2245 -0.2516 -0.0333 0.2052 0.3167 0.3388 0.2979 0.2408 0.2438 0.1288 0.0355 -0.0726 0.1884 0.5430 0.5037 0.3928 0.6536 0.9805 1.0402 0.8571 0.8218 0.8197 0.7281 0.5522 0.6925 1.0045 1.0707 0.6734 0.6085 1.0526 1.2845 1.1089 1.2953 1.3806 1.1912 0.9422 0.8630 0.9159 1.4254 1.8088 1.7759 1.2884 0.9880 0.8059 0.6631 0.3597 0.2877 0.5101 0.9324 1.1062 1.1189 1.0873 1.2963 1.4214 1.6282 1.7053 1.4973 0.9790 1.0983 1.3852 1.5429 1.5234 1.5292 1.3104 1.2757 0.8869 0.5145 0.7113 1.3543 1.3890 0.9667 0.7196 1.4537 2.0366 2.1264 1.8825 1.6793 1.3166 1.3493 1.3008 1.2872 1.2882 1.6478 1.8070 2.1904 2.2762 2.0932 1.6617 1.7195 2.0066 2.6213 2.6506 2.5683 2.3499 2.5807 2.5199 2.4837 1.9532 1.7694 1.8284 2.2359 2.1669 2.0885 1.9388 2.2269 2.5849 3.0879 2.8728 3.2147 3.0896 2.6566 1.8906 1.8568 1.8620 1.9197 0.5811 0.0707 -0.1140 -0.2507 -0.3872 -0.5261 -0.3324 0.6124 0.6609 0.7726 0.7911 0.8297 0.4739 0.2283 0.2148 0.4259 0.5007 -------------------------------