%PDF-1.3 1 0 obj > endobj 2 0 obj > endobj 3 0 obj > /ExtGState > >> /Font > /ProcSet [ /PDF /Text ] >> /MediaBox [ 0 0 612 792 ] /Annots [ 55 0 R 56 0 R 57 0 R 58 0 R ] >> endobj 4 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 90 0 R 91 0 R 92 0 R 93 0 R 94 0 R 95 0 R 96 0 R 97 0 R 98 0 R 99 0 R 100 0 R 101 0 R ] >> endobj 5 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 105 0 R 106 0 R 107 0 R 108 0 R 109 0 R 110 0 R 111 0 R 112 0 R 113 0 R 114 0 R 115 0 R 116 0 R 117 0 R 118 0 R 119 0 R 120 0 R 121 0 R 122 0 R 123 0 R 124 0 R 125 0 R 126 0 R 127 0 R 128 0 R 129 0 R 130 0 R 131 0 R 132 0 R 133 0 R 134 0 R 135 0 R 136 0 R 137 0 R 138 0 R 139 0 R 140 0 R 141 0 R 142 0 R 143 0 R 144 0 R 145 0 R 146 0 R 147 0 R 148 0 R 149 0 R ] >> endobj 6 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 181 0 R 182 0 R 183 0 R 184 0 R 185 0 R 186 0 R 187 0 R 188 0 R 189 0 R 190 0 R 191 0 R 192 0 R 193 0 R 194 0 R 195 0 R 196 0 R ] >> endobj 7 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 237 0 R 238 0 R 239 0 R 240 0 R 241 0 R 242 0 R ] >> endobj 8 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 264 0 R 265 0 R ] >> endobj 9 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 288 0 R 289 0 R 290 0 R 291 0 R 292 0 R 293 0 R ] >> endobj 10 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 297 0 R 298 0 R 299 0 R 300 0 R 301 0 R 302 0 R 303 0 R 304 0 R 305 0 R 306 0 R 307 0 R 308 0 R 309 0 R 310 0 R 311 0 R 312 0 R 313 0 R 314 0 R 315 0 R 316 0 R 317 0 R 318 0 R 319 0 R 320 0 R 321 0 R ] >> endobj 11 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 325 0 R 326 0 R 327 0 R 328 0 R 329 0 R 330 0 R 331 0 R 332 0 R 333 0 R 334 0 R ] >> endobj 12 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 363 0 R 364 0 R 365 0 R 366 0 R 367 0 R 368 0 R 369 0 R ] >> endobj 13 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 376 0 R 377 0 R 378 0 R 379 0 R 380 0 R 381 0 R 382 0 R 383 0 R 384 0 R 385 0 R 386 0 R 387 0 R 388 0 R 389 0 R ] >> endobj 14 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 420 0 R ] >> endobj 15 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 424 0 R 425 0 R 426 0 R 427 0 R 428 0 R 429 0 R 430 0 R 431 0 R 432 0 R 433 0 R 434 0 R 435 0 R 436 0 R 437 0 R 438 0 R 439 0 R 440 0 R 441 0 R ] >> endobj 16 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 445 0 R 446 0 R 447 0 R 448 0 R 449 0 R 450 0 R ] >> endobj 17 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 454 0 R 455 0 R 456 0 R 457 0 R 458 0 R 459 0 R 460 0 R 461 0 R 462 0 R 463 0 R 464 0 R 465 0 R 466 0 R 467 0 R 468 0 R 469 0 R 470 0 R 471 0 R 472 0 R 473 0 R 474 0 R 475 0 R 476 0 R 477 0 R 478 0 R 479 0 R 480 0 R 481 0 R 482 0 R 483 0 R 484 0 R 485 0 R ] >> endobj 18 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 489 0 R 490 0 R 491 0 R 492 0 R 493 0 R 494 0 R 495 0 R 496 0 R 497 0 R 498 0 R 499 0 R 500 0 R 501 0 R 502 0 R 503 0 R 504 0 R 505 0 R 506 0 R 507 0 R 508 0 R 509 0 R 510 0 R 511 0 R 512 0 R 513 0 R 514 0 R 515 0 R 516 0 R 517 0 R 518 0 R 519 0 R 520 0 R 521 0 R 522 0 R 523 0 R ] >> endobj 19 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 527 0 R 528 0 R 529 0 R 530 0 R 531 0 R 532 0 R 533 0 R 534 0 R 535 0 R 536 0 R 537 0 R 538 0 R 539 0 R 540 0 R 541 0 R 542 0 R 543 0 R 544 0 R 545 0 R 546 0 R 547 0 R 548 0 R 549 0 R 550 0 R 551 0 R 552 0 R 553 0 R 554 0 R 555 0 R 556 0 R 557 0 R 558 0 R ] >> endobj 20 0 obj > /MediaBox [ 0 0 612 792 ] /Annots [ 562 0 R 563 0 R 564 0 R 565 0 R 566 0 R 567 0 R 568 0 R 569 0 R 570 0 R 571 0 R 572 0 R 573 0 R 574 0 R 575 0 R ] >> endobj 21 0 obj > endobj 22 0 obj > stream q q 0.1 0 0 0.1 0 0 cm /R12 gs 0 g q 10 0 0 10 0 0 cm BT /R13 14.3462 Tf 1 0 0 1 205.276 666.202 Tm [ (A) 50.0017 (udio\055V) 37 (isual) -250.005 (Scene) -250.012 (Analysis) -250.003 (with) ] TJ -15.5652 -17.9328 Td [ (Self\055Super) 10 (vised) -249.99 (Multisensory) -250.008 (F) 24.9889 (eatur) 17.9912 (es) ] TJ /R15 9.9626 Tf 49.4891 -35.066 Td [ (Andre) 24.9811 (w) -249.995 (Owens) -1250.01 (Ale) 14.9828 (x) 14.9828 (ei) -249.995 (A\056) -249.993 (Efros) ] TJ /R15 8.9664 Tf 45.0582 -20.8801 Td [ (UC) -250.019 (Berk) 10 (ele) 15.0028 (y) ] TJ /R13 8.9664 Tf -121.147 -31.5188 Td (Abstract\056) Tj /R15 8.9664 Tf 40.5859 0 Td [ (The) -239.983 (thud) -239.994 (of) -241.018 (a) -240.021 (bouncing) -240.013 (ball\054) -239.98 (the) -241.015 (onset) -240.007 (of) -240.016 (speech) -240.01 (as) -240.016 (lips) -240.982 (open) -240.024 (\227) -240.002 (when) ] TJ -40.5859 -10.959 Td [ (visual) -364.988 (and) -365.01 (audio) -366.004 (e) 24.9875 (v) 14.9865 (ents) -364.997 (occur) -365 (together) 40.0012 (\054) -365.013 (it) -365.016 (suggests) -365.99 (t) 1.01017 (hat) -366.004 (there) -364.991 (might) -365.007 (be) -365.013 (a) -365.007 (com\055) ] TJ 10.959 TL T* [ (mon\054) -346.01 (underlying) -346.013 (e) 24.9875 (v) 14.9865 (ent) -347.009 (that) -345.997 (produced) -346.007 (both) -345.988 (signals\056) -346.01 (In) -347.012 (this) -346.018 (paper) 39.9849 (\054) -346.018 (we) -345.98 (ar) 18.0034 (gue) -346.018 (that) ] TJ T* [ (the) -381.991 (visual) -381.979 (and) -380.999 (audio) -381.992 (components) -381.998 (of) -381.995 (a) -382 (video) -380.991 (signal) -381.979 (should) -382.014 (be) -382 (modeled) -381.001 (jointly) ] TJ T* [ (using) -284.993 (a) -285.982 (fused) -284.991 (multisensory) -285.996 (representation\056) -285.013 (W) 80.0106 (e) -284.98 (propose) -285.993 (to) -284.997 (learn) -286.007 (such) -285.021 (a) -284.978 (represen\055) ] TJ T* [ (tation) -313.009 (in) -313.009 (a) -312.993 (self\055supervised) -311.996 (w) 10.0092 (ay) 64.9805 (\054) -312.996 (by) -313.018 (training) -312.979 (a) -312.993 (neural) -313.018 (netw) 10 (ork) -312.988 (to) -313.009 (predict) -312.005 (whether) ] TJ T* [ (video) -375.98 (frames) -376.98 (and) -375.989 (audio) -376.982 (are) -376.021 (temporally) -375.986 (aligned\056) -377.015 (W) 80.0106 (e) -375.989 (use) -376.985 (this) -375.991 (learned) -375.986 (represen\055) ] TJ T* [ (tation) -383.019 (for) -384.014 (three) -382.984 (applications\072) -383.994 (\050a\051) -382.986 (sound) -383.983 (source) -382.989 (localization\054) -383.994 (i\056e\056) -382.995 (visualizing) -382.984 (the) ] TJ T* [ (source) -333.018 (of) -333.987 (sound) -333.011 (in) -333.006 (a) -333.992 (vi) 1.01017 (deo\073) -333.983 (\050b\051) -332.998 (audio\055visual) -332.992 (action) -333.992 (recognition\073) -332.987 (and) -332.992 (\050c\051) -333.02 (on\057of) 25.0011 (f\055) ] TJ T* [ (screen) -377.997 (audio) -376.982 (source) -378.023 (separation\054) -376.999 (e\056g\056) -377.993 (remo) 15.0056 (ving) -378.009 (the) -376.982 (of) 24.9929 (f\055screen) -377.982 (translator\047) 55.0177 (s) -377.01 (v) 19.9965 (oice) ] TJ 10.9582 TL T* [ (from) -246.993 (a) -246.991 (foreign) -246.008 (of) 24.9929 55.0068 (s) -247.01 (speech\056) -247.002 (Code\054) -247 (models\054) -246.991 (and) -245.989 (video) -246.983 (results) -246.978 (are) -246.981 (a) 20.0238 (v) 25.0066 (ailable) -245.998 (on) ] TJ 10.959 TL T* [ (our) -249.993 (webpage\072) ] TJ ET Q 0 1 0 0 k q 10 0 0 10 0 0 cm BT /R15 8.9664 Tf 1 0 0 1 213.905 440.256 Tm [ (http\072\057\057andre) 24.9984 (w) 10.0092 (o) 25.0066 (wens\056com\057multisensory) ] TJ ET Q 0 g q 10 0 0 10 0 0 cm BT /R15 8.9664 Tf 1 0 0 1 350.598 440.256 Tm (\056) Tj /R13 11.9552 Tf -215.833 -26.3211 Td [ (1) -1000 (Intr) 18.0146 (oduction) ] TJ /R15 9.9626 Tf 20.9758 TL T* [ (As) -189.999 (humans\054) -190.015 (we) -190.009 (e) 15.0122 (xperience) -190.005 (our) -190 (w) 10 (orld) -189.993 (through) -190.017 (a) -190.012 (number) -189.988 (of) -190.005 (simultaneous) -189.988 (sensory) -189.988 (streams\056) ] TJ 11.9551 TL T* [ (When) -211.017 (we) -210.986 (bite) -211.989 (into) -211 (an) -210.984 (apple\054) -211.014 (not) -211.987 (only) -211.002 (do) -211.014 (we) -210.987 (taste) -211.992 (it\054) -211.006 (b) 20.0016 (ut) -211.011 (\227) -211.014 (as) -212.001 (Smith) -211.016 (and) -211.019 (Gasser) -210.979 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 449.703 381.004 Tm (1) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 454.684 381.004 Tm [ (\135) -210.989 (point) ] TJ -319.919 -11.9559 Td [ (out) -211.987 (\227) -211.994 (we) -212.005 (also) -211.01 (hear) -212.011 (it) -211.989 (crunch\054) -211.999 (see) -212.006 (its) -211.984 (red) -212.006 (skin\054) -212.016 (and) -211.019 (feel) -212.009 (the) -211.996 (coolness) -212.019 (of) -212.001 (its) -211.984 (core\056) -212.009 (The) -211.019 (coin\055) ] TJ 11.9551 TL T* [ (cidence) -269.014 (of) -269.013 (sensations) -267.995 (gi) 24.9885 (v) 14.9828 (es) -269.011 (us) -269.002 (strong) -268.989 (e) 25.0105 (vidence) -269.004 (that) -268.019 (the) 14.9852 (y) -269.009 (were) -268.985 (generated) -269.014 (by) -269.004 (a) -268 (common\054) ] TJ T* [ (underlying) -303.991 (e) 25.0105 (v) 14.9828 (ent) -303 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 208.478 345.138 Tm (2) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 213.459 345.138 Tm [ (\135\054) -303.988 (since) -302.998 (it) -304.013 (is) -304.015 (unlik) 9.98853 (ely) -303 (that) -304.013 (the) 14.9852 (y) -303.984 (co\055occurred) -303.016 (across) -303.984 (multiple) -303.003 (modal\055) ] TJ -78.6945 -11.9551 Td [ (ities) -272.99 (merely) -273.981 (by) -273.003 (chance\056) -273.998 (These) -273.006 (cross\055modal\054) -273.018 (temporal) -274.008 (co\055occurrences) -272.984 (therefore) -273.018 (pro) 14.9828 (vide) ] TJ 11.9551 TL T* [ (a) -264 (useful) -264.999 (learning) -264.017 (signal\072) -264.98 (a) -264 (model) -264.99 (that) -264.02 (is) -265.002 (trained) -263.983 (to) -265.002 (detect) -263.988 (them) -264.995 (ought) -264.007 (to) -265.005 (disco) 14.9975 (v) 14.9828 (er) -264.985 (mul) 1.00473 (ti\055) ] TJ T* [ (modal) -266.99 (structures) -266.013 (that) -267 (are) -266.987 (useful) -267 (for) -265.997 (other) -267.004 (tasks\056) -266.992 (In) -267.012 (much) -266.017 (of) -267.012 (traditional) -267.014 (computer) -266.01 (vision) ] TJ 11.9563 TL T* [ (research\054) -315.996 (ho) 24.986 (we) 25.0142 (v) 14.9828 (er) 39.986 (\054) -315.984 (we) -317.008 (ha) 19.9967 (v) 14.9828 (e) -315.991 (been) -315.984 (a) 19.9918 (v) 20.0016 (oiding) -315.991 (the) -316.018 (use) -316.999 (of) -315.984 (other) 40.0031 (\054) -315.984 (non\055visual) -315.986 (modalities\054) -316.016 (ar) 19.9869 (\055) ] TJ 11.9547 TL T* [ (guably) -250.013 (making) -250.006 (the) -249.99 (perception) -250.02 (problem) -250.01 (harder) 39.9884 (\054) -249.995 (not) -250.02 (easier) 55.008 (\056) ] TJ 14.9441 -12.3891 Td [ (In) -257.995 (this) -256.99 (paper) 39.9909 (\054) -257.994 (we) -257.999 (lear) 1.01454 (n) -257.991 (a) -258.001 (temporal\054) -256.989 (multisensory) -258.016 (representation) -258.011 (that) -257.001 (fuses) -257.986 (the) -257.006 (visual) ] TJ -14.9441 -11.9551 Td [ (and) -273.007 (audio) -272.995 (components) -273.011 (of) -273.011 (a) -273.018 (video) -272.996 (signal\056) -272.984 (W) 79.9866 (e) -273.018 (propose) -272.994 (to) -273.001 (train) -273.001 (this) -272.989 (model) -272.989 (without) -273.013 (using) ] TJ T* [ (an) 15.0183 (y) -279.007 (manually) -279.999 (labeled) -279 (data\056) -279.988 (That) -279 (is\054) -280.017 (rather) -279.015 (than) -280.019 (e) 15.0122 (xplicitly) -279.005 (telling) -279.997 (the) -279.005 (model) -280.007 (that\054) -278.993 (e\056g\056\054) -279.002 (it) ] TJ T* [ (should) -226.988 (associate) -225.987 (mo) 14.995 (ving) -226.999 (lips) -226.996 (with) -226.019 (speech) -227.018 (or) -227.018 (a) -226.006 (thud) -226.999 (with) -226.999 (a) -226.006 (bouncing) -226.989 (ball\054) -225.982 (we) -226.982 (ha) 19.9967 (v) 14.9828 (e) -226.987 (it) -225.987 (dis\055) ] TJ T* [ (co) 15.0183 (v) 14.9828 (er) -260.983 (these) -261.986 (audio\055visual) -261.013 (associations) -261.013 (through) -262.005 (self\055supervised) -260.986 (training) -260.986 (\133) ] TJ ET Q 0 1 0 rg q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 418.31 225.154 Tm (3) Tj ET Q 0 g q 10 0 0 10 0 0 cm BT /R15 9.9626 Tf 1 0 0 1 423.291 225.154 Tm [ (\135\056) -261.995 (S) 0.99493 (p) -1.01454 (e) 1.01454 64.999 (\054) ] TJ -288.527 -11.9551 Td [ (we) -253.998 (train) -253.985 (a) -254.002 (neural) -253.992 (netw) 10.0081 (ork) -253.009 (on) -253.987 (a) -254.002 (\223prete) 15.0122 (xt\224) -253.989 (task) -253.982 (of) -253.994 (detecting) -254.012 (misalignment) -254.012 (between) -253.012 (audio) ] TJ 11.9563 TL T* [ (and) -233.014 (visual) -232.994 (streams) -233 (in) -233.008 (synthetically\055shifted) -232.991 (videos\056) -232.993 (The) -233.015 (netw) 10.0081 (ork) -233.013 (observ) 14.9926 (es) -233.02 (ra) 15.0073 (w) -233.015 (audio) -233 (and) ] TJ 11.9547 TL T* [ (video) -233.982 (streams) -235 (\227) -233.99 (some) -235 (of) -233.998 (which) -233.985 (are) -234.993 (aligned\054) -234.017 (and) -235.015 (som) 1.00228 (e) -234.985 (that) -233.985 (ha) 19.9967 (v) 14.9828 (e) -234.985 (been) -234 (randomly) -234.01 (shifted) ] TJ T* [ (by) -253.007 (a) -254.002 (fe) 25.0081 (w) -253.014 (seconds) -254.018 (\227) -253.007 (and) -253.992 (we) -253.019 (task) -253.982 (it) -253.002 (with) -254.014 (distinguishing) -253.012 (between) -253.989 (the) -253.009 (tw) 10.0081 (o\056) -253.987 (This) -252.992 (turns) -252.997 (out) ] TJ T* [ (to) -203.993 (be) -204.004 (a) -204.01 (challenging) -204.01 (training) -204.015 (task) -203.99 (that) -203.99 (forces) -204.01 (the) -203.998 (netw) 10.0081 (ork) -203.998 (to) -203.993 (fuse) -204.003 (visual) -204.017 (motion) -204.005 (wi) 0.99003 (th) -203.995 (audio) ] TJ T* [ (information) -249.994 (and\054) -249.989 (in) -249.985 (the) -249.99 (process\054) -249.985 (learn) -249.997 (a) -250.002 (useful) -249.983 (audio\055visual) -249.995 (feature) -250.005 (representation\056) ] TJ 14.9441 -12.3891 Td [ (W) 79.9879 (e) -329.989 (demonstrate) -329.991 (the) -330.016 (usefulnes) 1.00228 (s) -330.018 (of) -329.982 (our) -330.016 (multisensory) -330.004 (representation) -329.999 (in) -330.014 (three) -329.004 (audio\055) ] TJ -14.9441 -11.9547 Td [ (visual) -310.981 (applications\072) -310.997 (\050a\051) -311.017 (sound) -312 (source) -311.002 (localization\054) -311.009 (\050b\051) -311.009 (audio\055visual) -311.004 (action) -312.012 (re) 1.00964 (cognition\073) ] TJ ET Q Q Q q q 1 1 1 rg /a0 gs 48.406 786.422 515.188 -52.699 re f q /s5 gs /x6 Do Q q /s7 gs /x8 Do Q q /s9 gs /x10 Do Q q /s11 gs /x12 Do Q Q Q endstream endobj 23 0 obj > >> /XObject > >> /Length 28 /Group > /BBox [ 123 744 519 776 ] /Type /XObject /Subtype /Form >> stream xœ+äÒO4PH/VЯ02QpÉç ä