{ "model": "gemini-3.8-flash", "experiment": "shaina_fresh_test", "tasks": { "t1": { "scores": { "n": 10, "accuracy": 0.4, "ci95": [ 0.1682, 0.6873 ], "balanced_accuracy": 0.4167, "cohen_kappa": 0.2405 }, "items": [ { "id": "t1_RVFA__seed_vc__153f__main__id00460_KCBOXmDu5Ks_00025__id08352", "clips": [ "RVFA__seed_vc__153f__main__id00460_KCBOXmDu5Ks_00025__id08352.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the video", "the audio" ], "gt_answer": "the audio", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVFA__seed_vc__77f__main__id01296_B7gb-0xbMn0_00024__id06068", "clips": [ "RVFA__seed_vc__77f__main__id01296_B7gb-0xbMn0_00024__id06068.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the audio", "both", "the video", "neither" ], "gt_answer": "the audio", "pred_answer": "the audio", "correct": true, "raw": "```json\n{\"answer\": \"the audio\"}\n```" }, { "id": "t1_RVRA__real__77f__vocoded__id05623_HbnUHzNq8XI_00003__id00739", "clips": [ "RVRA__real__77f__vocoded__id05623_HbnUHzNq8XI_00003__id00739.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "neither", "pred_answer": "neither", "correct": true, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVRA__real__153f__vocoded__id02959_hp-EccEnxCQ_00026__id08803", "clips": [ "RVRA__real__153f__vocoded__id02959_hp-EccEnxCQ_00026__id08803.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "neither", "pred_answer": "neither", "correct": true, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__s2v__153f__main__id03084_BQIuTs2QZUU_00021__id08748", "clips": [ "FVFA__s2v__153f__main__id03084_BQIuTs2QZUU_00021__id08748.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "both", "pred_answer": "the video", "correct": false, "raw": "```json\n{\"answer\": \"the video\"}\n```" }, { "id": "t1_FVFA__animate__77f__main__id05162_NpL77H2Shgk_00043__id03606", "clips": [ "FVFA__animate__77f__main__id05162_NpL77H2Shgk_00043__id03606.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the audio", "neither", "both", "the video" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__s2v__77f__main__id06390_17l4_o3k29c_00007__id07683", "clips": [ "FVFA__s2v__77f__main__id06390_17l4_o3k29c_00007__id07683.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_FVFA__facefusion__153f__main__id07786_bFZAH657cFs_00051__id07683", "clips": [ "FVFA__facefusion__153f__main__id07786_bFZAH657cFs_00051__id07683.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "both", "the audio", "the video" ], "gt_answer": "both", "pred_answer": "the audio", "correct": false, "raw": "```json\n{\"answer\": \"the audio\"}\n```" }, { "id": "t1_FVFA__facefusion__77f__main__id05670_8I_YYqRsihM_00041__id04900", "clips": [ "FVFA__facefusion__77f__main__id05670_8I_YYqRsihM_00041__id04900.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "the video", "both", "neither", "the audio" ], "gt_answer": "both", "pred_answer": "neither", "correct": false, "raw": "```json\n{\"answer\": \"neither\"}\n```" }, { "id": "t1_RVFA__seed_vc__77f__main__id01858_HBSDURviflg_00015__id01156", "clips": [ "RVFA__seed_vc__77f__main__id01858_HBSDURviflg_00015__id01156.mp4" ], "part": null, "question": "Which part of this audio-visual clip, if any, is AI-generated or manipulated? A real recording that was only processed, without changing who is seen or heard, still counts as real.", "options": [ "neither", "the audio", "both", "the video" ], "gt_answer": "the audio", "pred_answer": "the audio", "correct": true, "raw": "```json\n{\"answer\": \"the audio\"}\n```" } ] }, "t2": { "scores": { "n": 10, "accuracy": 0.2, "ci95": [ 0.0567, 0.5098 ], "balanced_accuracy": 0.25, "cohen_kappa": 0.0 }, "items": [ { "id": "t2_FVRA__s2v__77f__main__id08406_qr5lYg7ej40_00213__id05444", "clips": [ "FVRA__s2v__77f__main__id08406_qr5lYg7ej40_00213__id05444.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "full-body swap", "fully generated", "real", "face swap" ], "gt_answer": "fully generated", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__facefusion__77f__main__id08940_B5cvspWteMI_00009__id03946", "clips": [ "FVRA__facefusion__77f__main__id08940_B5cvspWteMI_00009__id03946.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "full-body swap", "fully generated", "real" ], "gt_answer": "face swap", "pred_answer": "face swap", "correct": true, "raw": "```json\n{\"answer\": \"face swap\"}\n```" }, { "id": "t2_RVRA__real__77f__main__id05444_QIxNYNGaVg0_00040__id08406", "clips": [ "RVRA__real__77f__main__id05444_QIxNYNGaVg0_00040__id08406.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "full-body swap", "face swap", "real", "fully generated" ], "gt_answer": "real", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_FVRA__animate__77f__main__id07054_gLIAqAQ40LQ_00053__id08610", "clips": [ "FVRA__animate__77f__main__id07054_gLIAqAQ40LQ_00053__id08610.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "fully generated", "real", "full-body swap", "face swap" ], "gt_answer": "full-body swap", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_RVRA__real__77f__main__id00761_C_ZBK3E8bJk_00027__id07616", "clips": [ "RVRA__real__77f__main__id00761_C_ZBK3E8bJk_00027__id07616.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "fully generated", "face swap", "full-body swap" ], "gt_answer": "real", "pred_answer": "real", "correct": true, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__facefusion__77f__main__id06480_YQD4AsloKcY_00076__id01534", "clips": [ "FVRA__facefusion__77f__main__id06480_YQD4AsloKcY_00076__id01534.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "face swap", "fully generated", "full-body swap" ], "gt_answer": "face swap", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" }, { "id": "t2_FVRA__s2v__77f__main__id03854_yfQGoMSBDx0_00151__id00509", "clips": [ "FVRA__s2v__77f__main__id03854_yfQGoMSBDx0_00151__id00509.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "full-body swap", "real", "fully generated" ], "gt_answer": "fully generated", "pred_answer": "face swap", "correct": false, "raw": "{\"answer\": \"face swap\"}" }, { "id": "t2_FVRA__animate__77f__main__id02959_h9Qwwv-DApg_00016__id03078", "clips": [ "FVRA__animate__77f__main__id02959_h9Qwwv-DApg_00016__id03078.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "face swap", "fully generated", "full-body swap", "real" ], "gt_answer": "full-body swap", "pred_answer": "face swap", "correct": false, "raw": "```json\n{\"answer\": \"face swap\"}\n```" }, { "id": "t2_FVRA__s2v__77f__main__id02359_hoPN5dCzSC8_00093__id06987", "clips": [ "FVRA__s2v__77f__main__id02359_hoPN5dCzSC8_00093__id06987.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "fully generated", "full-body swap", "real", "face swap" ], "gt_answer": "fully generated", "pred_answer": "real", "correct": false, "raw": "{\"answer\": \"real\"}" }, { "id": "t2_FVRA__animate__77f__main__id03681_Szvjxba6BTg_00066__id01664", "clips": [ "FVRA__animate__77f__main__id03681_Szvjxba6BTg_00066__id01664.mp4" ], "part": null, "question": "Which part of this video, if any, was generated or altered by AI? Ignore normal processing such as brightness, colour or compression.", "options": [ "real", "full-body swap", "face swap", "fully generated" ], "gt_answer": "full-body swap", "pred_answer": "real", "correct": false, "raw": "```json\n{\"answer\": \"real\"}\n```" } ] }, "t3": { "scores": { "n": 10, "accuracy": 0.5, "ci95": [ 0.2366, 0.7634 ], "balanced_accuracy": 0.3542, "cohen_kappa": 0.2647 }, "items": [ { "id": "t3_voice_A_RVRA__real__77f__vocoded__id00948_PCy1KCIsNxo_00024__id04278", "clips": [ "RVRA__real__77f__vocoded__id00948_PCy1KCIsNxo_00024__id04278.mp4", "RVRA__real__77f__main__id00948_z-P1KSBdNU8_00128__id08581.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of a different person", "fake of this person", "real, same person", "real, different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_face_A_RVRA__real__77f__main__id05877_q_ZAQf6KfTQ_00093__id05564", "clips": [ "RVRA__real__77f__main__id05877_q_ZAQf6KfTQ_00093__id05564.mp4", "RVRA__real__77f__main__id05877_6QUcq_6B-LU_00011__id07768.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, different person", "real, same person", "fake of this person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_voice_B_RVRA__real__153f__vocoded__id00885_xiPMwAYISG0_00135__id05623", "clips": [ "RVRA__real__153f__vocoded__id00885_xiPMwAYISG0_00135__id05623.mp4", "RVRA__real__153f__main__id05623_UXUHOpRirtg_00079__id03123.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "fake of this person", "real, different person", "fake of a different person" ], "gt_answer": "real, different person", "pred_answer": "real, same person", "correct": false, "raw": "```json\n{\"answer\": \"real, same person\"}\n```" }, { "id": "t3_face_B_RVRA__real__153f__main__id06713_5qTLKWiq73k_00074__id05042", "clips": [ "RVRA__real__153f__main__id06713_5qTLKWiq73k_00074__id05042.mp4", "RVRA__real__153f__main__id05042_K1HMS1K0kWI_00010__id00144.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, same person", "fake of a different person", "fake of this person", "real, different person" ], "gt_answer": "real, different person", "pred_answer": "real, different person", "correct": true, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_A_RVFA__seed_vc__153f__main__id04049_JCDeI7UwWnE_00080__id08096", "clips": [ "RVFA__seed_vc__153f__main__id04049_JCDeI7UwWnE_00080__id08096.mp4", "RVRA__real__153f__main__id04049_Flu_4soH0mo_00003__id04528.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of a different person", "real, same person", "real, different person", "fake of this person" ], "gt_answer": "fake of a different person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_B_RVFA__seed_vc__77f__main__id07574_DbcyMh1l_Ng_00043__id06429", "clips": [ "RVFA__seed_vc__77f__main__id07574_DbcyMh1l_Ng_00043__id06429.mp4", "RVRA__real__77f__main__id06429_0aTJ9yNMAx0_00002__id06427.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "fake of this person", "real, different person", "real, same person", "fake of a different person" ], "gt_answer": "fake of this person", "pred_answer": "fake of a different person", "correct": false, "raw": "```json\n{\"answer\": \"fake of a different person\"}\n```" }, { "id": "t3_face_A_RVRA__real__153f__main__id00885_yMWJySOEuTI_00138__id00739", "clips": [ "RVRA__real__153f__main__id00885_yMWJySOEuTI_00138__id00739.mp4", "RVRA__real__153f__main__id00885_xiPMwAYISG0_00135__id05623.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, different person", "fake of this person", "real, same person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_face_B_RVRA__real__77f__main__id08808_vQV6tYM_-3M_00053__id07697", "clips": [ "RVRA__real__77f__main__id08808_vQV6tYM_-3M_00053__id07697.mp4", "RVRA__real__77f__main__id07697_diNcpyVMCk8_00047__id07630.mp4" ], "part": "face", "question": "Clip 2 is a real recording of one person. Look only at the faces (the inner face: eyes, nose and mouth). Is the face in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose face it is, still counts as real.", "options": [ "real, same person", "fake of this person", "fake of a different person", "real, different person" ], "gt_answer": "real, different person", "pred_answer": "real, different person", "correct": true, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_B_RVFA__seed_vc__153f__main__id02359_QiCI0mefCx4_00030__id03078", "clips": [ "RVFA__seed_vc__153f__main__id02359_QiCI0mefCx4_00030__id03078.mp4", "RVRA__real__153f__main__id03078_4XN_hNJCe9E_00031__id06987.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "real, different person", "fake of a different person", "fake of this person" ], "gt_answer": "fake of this person", "pred_answer": "real, different person", "correct": false, "raw": "```json\n{\"answer\": \"real, different person\"}\n```" }, { "id": "t3_voice_A_RVRA__real__153f__vocoded__id01124_7Y-51ooBIWQ_00040__id09218", "clips": [ "RVRA__real__153f__vocoded__id01124_7Y-51ooBIWQ_00040__id09218.mp4", "RVRA__real__153f__main__id01124_2hg7q9O8xLo_00015__id08748.mp4" ], "part": "voice", "question": "Clip 2 is a real recording of one person. Listen only to the voices. Is the voice in clip 1 real or fake, and is it the same person as in clip 2? A real recording that was only processed, without changing whose voice it is, still counts as real.", "options": [ "real, same person", "fake of this person", "real, different person", "fake of a different person" ], "gt_answer": "real, same person", "pred_answer": "real, same person", "correct": true, "raw": "{\"answer\": \"real, same person\"}" } ] }, "t4": { "summary": { "n": 96, "with_reference": 84, "no_evidence": 0, "per_cell": { "FVRA_animate": 12, "FVFA_animate": 12, "FVFA_s2v": 12, "FVRA_facefusion": 12, "FVFA_facefusion": 12, "RVFA_real": 12, "FVRA_s2v": 12, "RVRA_real": 12 }, "top_artifacts": [ [ "metallic or buzzy timbre", 48 ], [ "voice-face mismatch", 40 ], [ "unnatural blur", 33 ], [ "texture jittering", 30 ], [ "blending seam", 27 ], [ "shape distortion", 24 ], [ "unnatural human movement", 21 ], [ "camera motion inconsistency", 6 ] ], "annotator": "gemini-3.6-flash", "version": "v5" }, "items": [ { "id": "t4_FVRA__animate__77f__vocoded__id04717_Yh8GcH-kL4A_00045__id00762", "clips": [ "FVRA__animate__77f__vocoded__id04717_Yh8GcH-kL4A_00045__id00762.mp4", "RVRA__real__77f__vocoded__id04717_Yh8GcH-kL4A_00045__id00762.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From start to end, the outline around the head and checkered shirt displays visible blurring against the background. Facial skin details show slight texture jittering as the head shifts position.", "audio": "The speaker's voice retains natural pitch contours and clear consonant articulation throughout the utterance." }, "artifacts": { "video": [ "blending seam", "texture jittering" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id08352_01tqy29EVE0_00002__id05457", "clips": [ "FVFA__animate__77f__main__id08352_01tqy29EVE0_00002__id05457.mp4", "RVRA__real__77f__vocoded__id08352_01tqy29EVE0_00002__id05457.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "At the start and middle of the clip, the outline along the hair and jawline displays unnatural blur against the background. As the speaker moves her mouth near the end, subtle texture jittering appears around her lips and lower cheeks.", "audio": "Throughout the clip, the voice carries a unnatural metallic timbre on spoken vowels and consonants. The deep vocal resonance also produces a noticeable voice-face mismatch relative to the speaker." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__77f__main__id09188_FCNUohJFn0k_00028__id03962", "clips": [ "FVFA__s2v__77f__main__id09188_FCNUohJFn0k_00028__id03962.mp4", "RVRA__real__77f__vocoded__id09188_FCNUohJFn0k_00028__id03962.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start through the middle of the clip, the head and hair exhibit a floaty, unnatural motion against a completely static background. When the mouth opens wide around the third and fourth frames, the lip contours show mild shape distortion.", "audio": "Throughout the clip, the vocal tone features a metallic, unnatural timbre. From start to finish, the voice character does not visually align with the speaker." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id06425_pCHzIeqRoXY_00041__id05096", "clips": [ "FVFA__s2v__153f__main__id06425_pCHzIeqRoXY_00041__id05096.mp4", "RVRA__real__153f__vocoded__id06425_pCHzIeqRoXY_00041__id05096.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "In the middle frames as the face speaks, the jaw and mouth undergo smooth shape distortion. The background and hair remain rigid and static while the face moves unnaturally.", "audio": "During the first phrase, the spoken voice carries a metallic, buzzy timbre. Near the end of the clip, the consonant transitions sound slightly smeared." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "smeared breaths or consonants" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id05457_Os6q_KNvHUU_00059__id06220", "clips": [ "FVRA__facefusion__77f__vocoded__id05457_Os6q_KNvHUU_00059__id06220.mp4", "RVRA__real__77f__vocoded__id05457_Os6q_KNvHUU_00059__id06220.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the middle of the clip, subtle skin tone mismatches appear along the jawline and cheeks. Towards the end, inner face details show slight blurring during mouth movement.", "audio": "The voice maintains natural pitch and clear cadence throughout the spoken phrase. Vocal tone and breath sounds remain consistent from start to end." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__77f__main__id03681_QmEgT37PFuY_00047__id06471", "clips": [ "FVFA__facefusion__77f__main__id03681_QmEgT37PFuY_00047__id06471.mp4", "RVRA__real__77f__vocoded__id03681_QmEgT37PFuY_00047__id06471.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A subtle seam is visible along the jawline and chin from the start through the middle frames where the inner face meets the neck and hand. The skin on the face looks unnaturally smooth compared to the neck area.", "audio": "A slightly buzzy sound is audible across the spoken phrase at the start and middle. The deep vocal timbre does not match the appearance of the speaker on screen." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__153f__main__id08871_99LJG_lch1E_00018__id02255", "clips": [ "RVFA__seed_vc__153f__main__id08871_99LJG_lch1E_00018__id02255.mp4", "RVRA__real__153f__vocoded__id08871_99LJG_lch1E_00018__id02255.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The facial movements, lip alignment, and lighting remain natural and consistent across all frames as the speaker tilts her head.", "audio": "During the phrase 'knowing you did the best you could', the speech has an unnatural metallic timbre and a noticeable voice-face mismatch." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__s2v__153f__vocoded__id06987_8Dr0Yc2qKhc_00018__id03078", "clips": [ "FVRA__s2v__153f__vocoded__id06987_8Dr0Yc2qKhc_00018__id03078.mp4", "RVRA__real__153f__vocoded__id06987_8Dr0Yc2qKhc_00018__id03078.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Near the start, the background behind the head remains completely static without natural camera movement. Around the middle during speech, the lip and chin region displays unnatural human movement and slight shape distortion.", "audio": "The speaker's voice maintains natural pitch transitions and clear vocal articulation from the beginning to the end of the sentence." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__153f__vocoded__id00657_6PAndSkJtWg_00021__id07818", "clips": [ "RVRA__real__153f__vocoded__id00657_6PAndSkJtWg_00021__id07818.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Facial expressions and mouth movement align naturally with speech around the center microphone frame from start to end. Head motions and lighting across the cheeks remain consistent.", "audio": "The speaker's voice retains natural cadence, pitch variations, and clear consonant sounds throughout the speech." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id08209_O8r8MtzKdfI_00036__id06429", "clips": [ "FVRA__facefusion__153f__vocoded__id08209_O8r8MtzKdfI_00036__id06429.mp4", "RVRA__real__153f__vocoded__id08209_O8r8MtzKdfI_00036__id06429.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "A faint color and texture mismatch is visible along the lower cheeks and jawline boundary as the woman speaks and turns her head slightly.", "audio": "The voice sounds natural with clear speech cadence, consistent pitch, and realistic vocal tone throughout." }, "artifacts": { "video": [ "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__s2v__77f__vocoded__id08581_jj2qWDyIC_8_00104__id03854", "clips": [ "FVRA__s2v__77f__vocoded__id08581_jj2qWDyIC_8_00104__id03854.mp4", "RVRA__real__77f__vocoded__id08581_jj2qWDyIC_8_00104__id03854.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The background and shoulders stay completely frozen without natural camera motion from start to end. Around the middle, the lower facial features morph unnaturally during speech while the surrounding scarf remains unnaturally motion-free.", "audio": "The spoken voice maintains a natural timbre and smooth pitch contours from the opening phrase through the end of the clip." }, "artifacts": { "video": [ "unnatural human movement", "camera motion inconsistency" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__153f__main__id00509_cJGftjdLydk_00036__id00948", "clips": [ "FVFA__facefusion__153f__main__id00509_cJGftjdLydk_00036__id00948.mp4", "RVRA__real__153f__vocoded__id00509_cJGftjdLydk_00036__id00948.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Throughout the clip, the skin tone on the inner face is smoother and slightly different in tone compared to the forehead and neck. Around the middle when the speaker talks, a subtle seam is visible near the top of the forehead and along the cheeks.", "audio": "A slightly buzzy, synthetic timbre is audible during the speech from the start to the middle of the clip. The vocal voice tone also shows a slight mismatch with the visual appearance of the speaker when pronouncing the word Dubai at the end." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__77f__vocoded__id05623_UXUHOpRirtg_00069__id03123", "clips": [ "RVRA__real__77f__vocoded__id05623_UXUHOpRirtg_00069__id03123.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Near the start, the mouth moves fluidly with natural lip shapes around the teeth and chin. Through the middle and end, facial lighting and skin texture remain consistent across the face.", "audio": "From start to end, the voice exhibits natural pitch variations and standard vocal inflection. Words and breath sounds transition smoothly throughout the spoken phrase." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__77f__main__id04126_ug07wiareEc_00220__id06015", "clips": [ "RVFA__seed_vc__77f__main__id04126_ug07wiareEc_00220__id06015.mp4", "RVRA__real__77f__vocoded__id04126_ug07wiareEc_00220__id06015.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The speaker's facial movements and lighting remain natural from start to end.", "audio": "The voice exhibits an unnatural buzzy timbre and a voice-face mismatch throughout the spoken phrase." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id03613_y-weegLHBmA_00147__id00476", "clips": [ "FVFA__animate__77f__main__id03613_y-weegLHBmA_00147__id00476.mp4", "RVRA__real__77f__vocoded__id03613_y-weegLHBmA_00147__id00476.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From 00:00 to 00:03, the skin texture on the man's cheeks and around his beard appears unnaturally smooth and blurred. Minor flickering occurs along his right jawline as his head subtly moves.", "audio": "The voice has a slightly buzzy and metallic quality during speech. Additionally, the vocal timbre creates an audible mismatch with the person displayed." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id00024_3ysHAigPe68_00005__id00330", "clips": [ "FVRA__animate__77f__vocoded__id00024_3ysHAigPe68_00005__id00330.mp4", "RVRA__real__77f__vocoded__id00024_3ysHAigPe68_00005__id00330.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "When the head turns near the middle of the clip, the hair outline against the background exhibits soft blurring and edge jittering.", "audio": "The voice sounds natural with realistic pitch variations and clear articulation throughout the spoken sentence." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id03078_w4vPdwaJj8A_00189__id02359", "clips": [ "FVFA__animate__77f__main__id03078_w4vPdwaJj8A_00189__id02359.mp4", "RVRA__real__77f__vocoded__id03078_w4vPdwaJj8A_00189__id02359.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Around the dreadlocks and white ruffle collar, soft blur and texture jittering occur as the head shifts throughout the clip. The lower edge of the collar exhibits mild shape distortion when the chin moves near the middle frames.", "audio": "A faint buzzy timbre is audible during the phrase 'of UEFA' at the start of the clip. The vocal tone presents a subtle voice-face mismatch relative to the visual speaker from start to end." }, "artifacts": { "video": [ "unnatural blur", "texture jittering", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id01708_PPlm6v3lrhM_00021__id02472", "clips": [ "FVRA__animate__77f__vocoded__id01708_PPlm6v3lrhM_00021__id02472.mp4", "RVRA__real__77f__vocoded__id01708_PPlm6v3lrhM_00021__id02472.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the head and neck, the hair edges softly blur into the background near the middle of the clip. The facial features and hair outline show mild texture jittering as she speaks toward the end.", "audio": "The voice demonstrates natural pitch contours, clean consonant articulation, and realistic speech cadence during her response." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id01348_tJpEZlLI7Do_00054__id03962", "clips": [ "FVRA__facefusion__77f__vocoded__id01348_tJpEZlLI7Do_00054__id03962.mp4", "RVRA__real__77f__vocoded__id01348_tJpEZlLI7Do_00054__id03962.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "A subtle blending seam and skin texture shift appear along the lower cheeks and jawline from the start through the end of the clip.", "audio": "The speech maintains a natural vocal tone with clear pronunciation and realistic breathing throughout the clip." }, "artifacts": { "video": [ "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__77f__main__id07768_U2QOxY66bgU_00081__id05564", "clips": [ "FVFA__facefusion__77f__main__id07768_U2QOxY66bgU_00081__id05564.mp4", "RVRA__real__77f__vocoded__id07768_U2QOxY66bgU_00081__id05564.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Around the middle of the clip, a subtle line of differing skin tone is visible where the inner cheeks meet the beard line. When the man speaks during the first half, the inner facial skin texture appears smoother than the surrounding neck.", "audio": "At the start when the speaker says 'I mean rose', the voice carries a buzzy and artificial timbre. During the middle frames, the audio tone sounds detached from the speaker's visual appearance." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__153f__vocoded__id03281_C7Y5tyjabNY_00050__id04528", "clips": [ "FVRA__s2v__153f__vocoded__id03281_C7Y5tyjabNY_00050__id04528.mp4", "RVRA__real__153f__vocoded__id03281_C7Y5tyjabNY_00050__id04528.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, the background behind the head remains completely static while the face moves. Around the middle of the clip, the mouth and jaw display unnatural warping as the subject speaks. Near the end, the inner mouth and teeth show unnatural blurring during speech movements.", "audio": "The voice maintains a natural timbre, smooth intonation, and clear consonant sounds throughout the clip." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__s2v__77f__vocoded__id08502_EILg7yn3EgU_00040__id04234", "clips": [ "FVRA__s2v__77f__vocoded__id08502_EILg7yn3EgU_00040__id04234.mp4", "RVRA__real__77f__vocoded__id08502_EILg7yn3EgU_00040__id04234.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the first to the final frame, the background behind the subject remains unnaturally static while the face warps subtly. Across frames 1 through 16, the speaker's head and neck exhibit floating, unnatural motion.", "audio": "The speech maintains natural pitch variation and crisp consonant sounds from start to finish." }, "artifacts": { "video": [ "shape distortion", "unnatural human movement" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id07430_jJWKEnq3D-E_00127__id01156", "clips": [ "FVRA__facefusion__153f__vocoded__id07430_jJWKEnq3D-E_00127__id01156.mp4", "RVRA__real__153f__vocoded__id07430_jJWKEnq3D-E_00127__id01156.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Along the lower jaw and chin region around frames 5 through 8, a faint edge seam and skin texture mismatch distinguish the inner face from the surrounding neck. The interior facial features appear slightly softer in detail than the outer hairline.", "audio": "The audio features natural speech pacing, smooth tonal variations, and clear vocal articulation throughout the clip." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__153f__vocoded__id04234_IxLbObPoYCY_00048__id06480", "clips": [ "RVRA__real__153f__vocoded__id04234_IxLbObPoYCY_00048__id06480.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "At the start, facial expressions and smiling appear natural. As the speaker turns and speaks through the middle and end, lip movements and facial details remain coherent with the background.", "audio": "The voice maintains natural pitch contours and clean speech articulation from start to finish." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__77f__main__id07054_uFcEE5wJrOU_00065__id04236", "clips": [ "RVFA__seed_vc__77f__main__id07054_uFcEE5wJrOU_00065__id04236.mp4", "RVRA__real__77f__vocoded__id07054_uFcEE5wJrOU_00065__id04236.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Facial expressions and mouth movements appear natural throughout the speech. Lighting on the face and head remains consistent from start to end.", "audio": "The vocal timbre sounds slightly buzzy and metallic during the speech. There is also a noticeable voice-face mismatch with the speaker." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id00144_myrnp6zXitU_00153__id05042", "clips": [ "FVFA__s2v__153f__main__id00144_myrnp6zXitU_00153__id05042.mp4", "RVRA__real__153f__vocoded__id00144_myrnp6zXitU_00153__id05042.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "When the mouth puckers near the start around the fourth frame, the lips exhibit shape distortion while the left earring jitters. From the middle to the end, the dark background behind the blonde hair remains unnaturally frozen during head movements.", "audio": "Around the middle of the clip during the phrase 'such admiration', the vocal timbre carries a faint metallic buzz. Near the end on the word 'community', the pitch transitions sound unnaturally flat." }, "artifacts": { "video": [ "shape distortion", "texture jittering", "camera motion inconsistency" ], "audio": [ "metallic or buzzy timbre", "unnatural pitch or loudness" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__77f__vocoded__id04717_7DiEgpK7dbk_00021__id00761", "clips": [ "RVRA__real__77f__vocoded__id04717_7DiEgpK7dbk_00021__id00761.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The facial movements and skin textures around the mouth and eyes remain consistent and natural across all frames from start to finish.", "audio": "The audio features natural cadence, clear articulation, and steady vocal tone that align smoothly with the speaker's lip movement throughout the clip." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__77f__main__id07786_F29M4B5xDQ4_00026__id07839", "clips": [ "FVFA__s2v__77f__main__id07786_F29M4B5xDQ4_00026__id07839.mp4", "RVRA__real__77f__vocoded__id07786_F29M4B5xDQ4_00026__id07839.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Around the third frame as the mouth opens, the lips and inner mouth suffer shape distortion into a smoothed white shape. Near the middle of the clip, subtle texture jittering occurs along the lower chin during speech.", "audio": "At the start of the clip during the opening spoken phrase, a faint metallic timbre affects the vocal tone. The resulting voice profile creates a mild voice-face mismatch relative to the person shown." }, "artifacts": { "video": [ "shape distortion", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__153f__main__id00948_Qppg8ZCTQLE_00038__id00509", "clips": [ "RVFA__seed_vc__153f__main__id00948_Qppg8ZCTQLE_00038__id00509.mp4", "RVRA__real__153f__vocoded__id00948_Qppg8ZCTQLE_00038__id00509.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The speaker's facial expressions and lip movements align naturally with his speech across all frames. The lighting on his face and the background banner remain consistent throughout the clip.", "audio": "At 00:01 on the word 'iconic', the voice has a buzzy, unnatural texture. The vocal timbre also exhibits a noticeable voice-face mismatch relative to the young man shown." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__153f__main__id04234_P4SBvgQH1OI_00077__id01534", "clips": [ "FVFA__facefusion__153f__main__id04234_P4SBvgQH1OI_00077__id01534.mp4", "RVRA__real__153f__vocoded__id04234_P4SBvgQH1OI_00077__id01534.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A subtle line is visible along the cheeks and jawline where the inner face joins the head structure. During mouth movement in the middle of the clip, minor texture jittering appears around the mustache.", "audio": "The speech exhibits a slight metallic quality throughout the audio track. Additionally, the vocal timbre presents a noticeable mismatch with the speaker's visual age." }, "artifacts": { "video": [ "blending seam", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__animate__77f__vocoded__id08803_SoxImR0IXkw_00058__id02359", "clips": [ "FVRA__animate__77f__vocoded__id08803_SoxImR0IXkw_00058__id02359.mp4", "RVRA__real__77f__vocoded__id08803_SoxImR0IXkw_00058__id02359.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The mouth movements appear slightly stiff and misaligned with the spoken words as the subject speaks. Subtle blurring is visible along the edges of the long hair where it meets the background.", "audio": "The speech sounds natural and clear with well-defined consonants and consistent vocal cadence throughout." }, "artifacts": { "video": [ "lip-sync mismatch", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id07630_Q4FWh9vQCvc_00021__id07697", "clips": [ "FVFA__animate__77f__main__id07630_Q4FWh9vQCvc_00021__id07697.mp4", "RVRA__real__77f__vocoded__id07630_Q4FWh9vQCvc_00021__id07697.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "In the middle frames as the face moves, an unnatural blur appears along the outer edge of the left cheek and hair. Subtle texture jittering occurs around the mouth and chin during speech in the second half of the clip.", "audio": "A faint metallic timbre is present during the vocal sound at the start. Near the end of the clip, consonants sound slightly smeared as the speaker finishes talking." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "smeared breaths or consonants" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id07697_diNcpyVMCk8_00065__id07630", "clips": [ "FVRA__animate__77f__vocoded__id07697_diNcpyVMCk8_00065__id07630.mp4", "RVRA__real__77f__vocoded__id07697_diNcpyVMCk8_00065__id07630.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Along the hair and neck boundary, subtle edge jitter occurs against the dark backdrop. As the speaker mouth moves around the middle frames, minor structural distortion is visible around the lips.", "audio": "The speech pitch and phrasing remain natural, with clear consonant articulation and smooth vocal cadence from start to end." }, "artifacts": { "video": [ "texture jittering", "structure anomaly" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id08790_UtGvVIJVqgg_00119__id06429", "clips": [ "FVFA__animate__77f__main__id08790_UtGvVIJVqgg_00119__id06429.mp4", "RVRA__real__77f__vocoded__id08790_UtGvVIJVqgg_00119__id06429.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "At the start and middle of the clip, the facial features display unnatural blur and smoothing. Near the end, subtle texture jittering appears along the hair outline against the dark background.", "audio": "During the spoken phrase from the start, the acoustic texture has a faint buzzy timbre. The voice characteristics also show a slight voice-face mismatch with the speaker." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__153f__main__id03479_wKMnsqfVAcI_00187__id00330", "clips": [ "FVFA__facefusion__153f__main__id03479_wKMnsqfVAcI_00187__id00330.mp4", "RVRA__real__153f__vocoded__id03479_wKMnsqfVAcI_00187__id00330.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A soft blending boundary and slight blurring are visible along the cheeks and jawline during the middle frames as the speaker talks.", "audio": "The vocal audio displays a faint metallic timbre with slightly smeared consonant sounds throughout the speech." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "smeared breaths or consonants" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__77f__vocoded__id08338_bhL_BGw8zsI_00037__id01643", "clips": [ "RVRA__real__77f__vocoded__id08338_bhL_BGw8zsI_00037__id01643.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The speaker's facial expressions, mouth movements, and eye blinks transition naturally from the start to the end of the clip. Lighting and skin textures on the face remain consistently lifelike throughout.", "audio": "The German speech maintains a steady, natural pitch and clear articulation from the start to the end. Vocal timbre and natural breath sounds remain consistent throughout." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__153f__vocoded__id04709_TdSjp6WaNSI_00112__id01124", "clips": [ "FVRA__s2v__153f__vocoded__id04709_TdSjp6WaNSI_00112__id01124.mp4", "RVRA__real__153f__vocoded__id04709_TdSjp6WaNSI_00112__id01124.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, the background backdrop remains unnaturally static with no organic camera motion or micro-movements. During the middle frames, the facial expressions and lip movements show synthetic warping rather than natural jaw articulation.", "audio": "The voice maintains a natural cadence and clear pronunciation on words like 'money' and 'education' from start to end. The pitch contours and phrasing sound smooth and natural throughout the speech." }, "artifacts": { "video": [ "camera motion inconsistency", "unnatural human movement" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id08338_bhL_BGw8zsI_00038__id01643", "clips": [ "FVRA__facefusion__153f__vocoded__id08338_bhL_BGw8zsI_00038__id01643.mp4", "RVRA__real__153f__vocoded__id08338_bhL_BGw8zsI_00038__id01643.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "At the start and middle of the clip, a boundary line and slight skin tone difference appear along the upper cheeks and jawline. The inner facial area also exhibits softer lighting compared to the surrounding neck and ears.", "audio": "The speech maintains a natural cadence and clear acoustic quality from the start to the end of the statement." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__77f__main__id02959_hp-EccEnxCQ_00022__id08803", "clips": [ "FVFA__facefusion__77f__main__id02959_hp-EccEnxCQ_00022__id08803.mp4", "RVRA__real__77f__vocoded__id02959_hp-EccEnxCQ_00022__id08803.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A faint edge along the lower cheeks and chin shows a subtle texture shift near the middle of the clip. Unnatural smoothing appears around the lips and mouth during speech toward the end.", "audio": "The voice carries a persistent metallic texture throughout the speech. Consonant sounds lack crispness when uttering the final phrase at the end of the clip." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "smeared breaths or consonants" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__77f__vocoded__id04900_yQW2qx3Yq0w_00034__id05162", "clips": [ "FVRA__s2v__77f__vocoded__id04900_yQW2qx3Yq0w_00034__id05162.mp4", "RVRA__real__77f__vocoded__id04900_yQW2qx3Yq0w_00034__id05162.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the middle to the end of the clip, the mouth and lower lip distort unnaturally while speaking. The head and shoulders remain rigidly static without natural body movement against a completely frozen background.", "audio": "The voice maintains a consistent, natural timbre and smooth speech cadence during the spoken phrase from start to end." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion", "lip-sync mismatch" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__153f__main__id00024_IJ3QWgaZC-g_00036__id03479", "clips": [ "RVFA__seed_vc__153f__main__id00024_IJ3QWgaZC-g_00036__id03479.mp4", "RVRA__real__153f__vocoded__id00024_IJ3QWgaZC-g_00036__id03479.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The speaker's facial expressions and lip movements remain natural from the start to the end. Her skin texture and facial lighting stay consistent throughout the recording.", "audio": "The voice exhibits a slightly buzzy, artificial timbre on the words 'first round' and '50 girls'. The resulting vocal tone creates a distinct voice-face mismatch relative to the speaker." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__77f__main__id00762_jo0CcA_tDl8_00151__id00761", "clips": [ "RVFA__seed_vc__77f__main__id00762_jo0CcA_tDl8_00151__id00761.mp4", "RVRA__real__77f__vocoded__id00762_jo0CcA_tDl8_00151__id00761.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The speaker's facial expressions, mouth movements, and skin textures look completely natural as he speaks from start to end. Lighting on his face and background remains consistent throughout the clip.", "audio": "The voice has an artificial, buzzy quality throughout the speech from 00:00 to 00:03. The voice timbre also creates a noticeable mismatch with the speaker's appearance on screen." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id00024_IJ3QWgaZC-g_00034__id03479", "clips": [ "FVFA__s2v__153f__main__id00024_IJ3QWgaZC-g_00034__id03479.mp4", "RVRA__real__153f__vocoded__id00024_IJ3QWgaZC-g_00034__id03479.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The background behind the head warps and distorts unnaturally as the face moves from the start to the end. Near the beginning when speaking 'strict', the mouth shape and cheek texture display unnatural human movement and slight distortion.", "audio": "The audio has a buzzy, metallic timbre noticeable on words like 'strict' near the start. The vocal acoustic characteristics also create a voice-face mismatch with the speaker." }, "artifacts": { "video": [ "shape distortion", "texture jittering", "unnatural human movement" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id07839__QTWWQBdhWY_00029__id07683", "clips": [ "FVRA__facefusion__77f__vocoded__id07839__QTWWQBdhWY_00029__id07683.mp4", "RVRA__real__77f__vocoded__id07839__QTWWQBdhWY_00029__id07683.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end of the clip, the inner face features display a distinct facial identity that differs from the outer head structure. Slight color boundary differences appear along the cheek edges as the face moves near the middle.", "audio": "The speaker's voice sounds natural with clear consonant pronunciation and steady tone from the beginning to the end of the clip." }, "artifacts": { "video": [ "person identity inconsistency", "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__153f__vocoded__id08406_qr5lYg7ej40_00216__id05444", "clips": [ "RVRA__real__153f__vocoded__id08406_qr5lYg7ej40_00216__id05444.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Facial movements, eye blinks, and lip motion appear consistent and natural as the woman speaks from the start to the end of the clip. Lighting on her face and background remains stable throughout.", "audio": "The speech delivery sounds natural with clear speech cadence and realistic vocal resonance from start to finish." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__s2v__77f__main__id08170_RAqiEPNyzgk_00015__id04414", "clips": [ "FVFA__s2v__77f__main__id08170_RAqiEPNyzgk_00015__id04414.mp4", "RVRA__real__77f__vocoded__id08170_RAqiEPNyzgk_00015__id04414.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start to the middle of the clip, the subject's head and hair exhibit subtle morphing while the surrounding background stays completely static. The lower facial movements display mild shape distortion around the mouth and lips while speaking.", "audio": "Throughout the clip, the spoken voice carries a slight metallic timbre and muffled consonant clarity. The audio tone presents minor unnatural pitch shifts during spoken phrases." }, "artifacts": { "video": [ "shape distortion", "unnatural human movement" ], "audio": [ "metallic or buzzy timbre", "unnatural pitch or loudness" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id07818_8CBP1GZl_7g_00030__id04052", "clips": [ "FVFA__animate__77f__main__id07818_8CBP1GZl_7g_00030__id04052.mp4", "RVRA__real__77f__vocoded__id07818_8CBP1GZl_7g_00030__id04052.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Near the start, subtle texture jittering appears around the hair outlines against the background. Around the middle, mild unnatural blur is visible around the mouth during speech movements.", "audio": "Near the end on 'contextually', the voice has a slightly metallic or buzzy timbre. The acoustic pitch also exhibits a mild voice-face mismatch relative to the visual speaker." }, "artifacts": { "video": [ "texture jittering", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id00330_d1jH7teHNr0_00078__id00024", "clips": [ "FVRA__animate__77f__vocoded__id00330_d1jH7teHNr0_00078__id00024.mp4", "RVRA__real__77f__vocoded__id00330_d1jH7teHNr0_00078__id00024.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "In the middle frames, subtle texture jittering and shape distortion appear around the hair edges and jawline against the background.", "audio": "The speech maintains natural pitch, clear consonant articulation, and consistent voice tone throughout the clip." }, "artifacts": { "video": [ "texture jittering", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id04126_vQ9igU1IY2w_00231__id03710", "clips": [ "FVRA__animate__77f__vocoded__id04126_vQ9igU1IY2w_00231__id03710.mp4", "RVRA__real__77f__vocoded__id04126_vQ9igU1IY2w_00231__id03710.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the speaker's hair and shoulders at the start and middle of the clip, soft edge blurring is visible against the background. The facial skin retains an unnaturally smooth texture throughout the sequence.", "audio": "The spoken voice sounds natural with clear articulation and consistent tone from start to finish." }, "artifacts": { "video": [ "unnatural blur", "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id05623_HbnUHzNq8XI_00003__id00739", "clips": [ "FVFA__animate__77f__main__id05623_HbnUHzNq8XI_00003__id00739.mp4", "RVRA__real__77f__vocoded__id05623_HbnUHzNq8XI_00003__id00739.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Unnatural blur appears around the hair boundaries and face edges against the background. Minor texture jittering is noticeable around the mouth area while speaking.", "audio": "The speaker's voice timbre is mismatched with the visual persona presented in the frames. A faint metallic quality persists throughout the spoken phrase." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "voice-face mismatch", "metallic or buzzy timbre" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__77f__main__id08581_Ex7FQ6ecyZc_00052__id00509", "clips": [ "FVFA__s2v__77f__main__id08581_Ex7FQ6ecyZc_00052__id00509.mp4", "RVRA__real__77f__vocoded__id08581_Ex7FQ6ecyZc_00052__id00509.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "In the middle of the clip, subtle texture jittering appears around the lips as the mouth moves. The background behind the speaker remains completely static without natural camera depth.", "audio": "Throughout the spoken phrase, the voice carries a slightly metallic timbre. The overall tone creates a noticeable voice-face mismatch with the person on screen." }, "artifacts": { "video": [ "texture jittering", "unnatural human movement" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__77f__main__id07574_7LUR89fORKY_00023__id08209", "clips": [ "FVFA__facefusion__77f__main__id07574_7LUR89fORKY_00023__id08209.mp4", "RVRA__real__77f__vocoded__id07574_7LUR89fORKY_00023__id08209.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A subtle boundary line with mismatched skin texture is visible along the cheek and jawline edges throughout the clip. Around the middle, the motion of the lips and teeth appears slightly blurred compared to the rest of the face.", "audio": "The spoken audio carries a faint metallic buzz across the dialogue. Additionally, the voice timbre sounds unnatural and mismatched with the speaker's visual profile." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__77f__vocoded__id06758_anvb4r0RKwU_00112__id01708", "clips": [ "RVRA__real__77f__vocoded__id06758_anvb4r0RKwU_00112__id01708.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The facial expressions and lip movements around the mouth adapt naturally as the woman speaks from the start to the middle of the clip. Her hair and background lighting maintain a consistent appearance through the final frames.", "audio": "The female Spanish speech maintains a natural pitch and clear inflection from the start through the end of the clip." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__153f__vocoded__id04900_NSha4VhuPh8_00010__id04901", "clips": [ "RVRA__real__153f__vocoded__id04900_NSha4VhuPh8_00010__id04901.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "At the start and middle, the subject's subtle head motions and lip shapes align naturally with speech. Lighting on her face and hair remains consistent as she looks slightly downward near the end.", "audio": "The voice maintains a natural cadence and clear tone while stating 'tightening the bonds' from start to finish. Speech pauses and vocal articulation sound well-defined throughout the clip." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id00165_6ildPS_kGwk_00012__id08667", "clips": [ "FVRA__facefusion__153f__vocoded__id00165_6ildPS_kGwk_00012__id08667.mp4", "RVRA__real__153f__vocoded__id00165_6ildPS_kGwk_00012__id08667.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Along the upper forehead and eyebrow boundary, a skin tone mismatch and soft edge are visible throughout the clip relative to the outer head.", "audio": "The spoken speech maintains natural pitch variations, clear consonant articulation, and consistent room acoustics from start to end." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__77f__main__id03123_3GdossNwwVs_00017__id05623", "clips": [ "RVFA__seed_vc__77f__main__id03123_3GdossNwwVs_00017__id05623.mp4", "RVRA__real__77f__vocoded__id03123_3GdossNwwVs_00017__id05623.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The speaker's face, curly hair, and background text move naturally with clear edge definition when she turns her head slightly at the start.", "audio": "The vocal timbre sounds slightly buzzy and muffled on the word 'standard' near the start. The voice texture creates a subtle mismatch with the speaker seen in the frame." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id08790_UtGvVIJVqgg_00121__id06429", "clips": [ "FVFA__s2v__153f__main__id08790_UtGvVIJVqgg_00121__id06429.mp4", "RVRA__real__153f__vocoded__id08790_UtGvVIJVqgg_00121__id06429.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "At the start and middle of the clip, the subject's mouth motion shows shape distortion and unnatural blurring around her teeth. Subtle texture jittering appears along the edges of her hair as her head moves.", "audio": "From the start to the end of the clip, the audio presents a distinct voice-face mismatch relative to the visual speaker. A slight metallic timbre is audible throughout her spoken words." }, "artifacts": { "video": [ "shape distortion", "unnatural blur", "texture jittering" ], "audio": [ "voice-face mismatch", "metallic or buzzy timbre" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__s2v__77f__vocoded__id05768_omj0fpqbVU4_00105__id07839", "clips": [ "FVRA__s2v__77f__vocoded__id05768_omj0fpqbVU4_00105__id07839.mp4", "RVRA__real__77f__vocoded__id05768_omj0fpqbVU4_00105__id07839.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the middle to the end of the clip, the lower lip and mouth motion display unnatural, rigid warping during speech. The background and camera angle remain unnaturally static from start to finish.", "audio": "The spoken voice demonstrates natural pitch contours, distinct cadence, and realistic articulation from start to end." }, "artifacts": { "video": [ "unnatural human movement", "camera motion inconsistency" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id07786_VwKW5Tj_I5U_00041__id06390", "clips": [ "FVRA__facefusion__77f__vocoded__id07786_VwKW5Tj_I5U_00041__id06390.mp4", "RVRA__real__77f__vocoded__id07786_VwKW5Tj_I5U_00041__id06390.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the middle of the clip, a light color boundary is visible along the right cheek near the hair. In the second half of the clip, the inner face texture appears smoother than the surrounding forehead as the mouth moves.", "audio": "The speaker's voice features natural cadence and pitch variations from the start of the clip. Consonants and spoken words remain clear and well-defined through the middle and final frames." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__153f__vocoded__id05441_df8dE7alJ-g_00066__id03962", "clips": [ "FVRA__s2v__153f__vocoded__id05441_df8dE7alJ-g_00066__id03962.mp4", "RVRA__real__153f__vocoded__id05441_df8dE7alJ-g_00066__id03962.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, the head exhibits floating, unnatural motion while the background remains entirely rigid. Around the middle frames while speaking, subtle texture jittering and shape distortion occur around the lips and chin.", "audio": "The speaker's voice carries natural inflection, smooth phrasing, and consistent vocal pitch from the beginning to the end of the clip." }, "artifacts": { "video": [ "unnatural human movement", "texture jittering", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__153f__main__id03123_Bd9QDl0lQ_k_00045__id00739", "clips": [ "RVFA__seed_vc__153f__main__id03123_Bd9QDl0lQ_k_00045__id00739.mp4", "RVRA__real__153f__vocoded__id03123_Bd9QDl0lQ_k_00045__id00739.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The facial movements and lip alignment progress naturally as the woman speaks from the start to the end. Lighting on her face and the background remains consistent across the entire clip.", "audio": "Around the middle of the clip when saying 'sostenibile', the vocal texture sounds slightly buzzy and unnatural. The voice quality also shows a slight mismatch with the visual appearance of the speaker." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__153f__main__id04278_ML08p1RpYQ4_00043__id03854", "clips": [ "FVFA__facefusion__153f__main__id04278_ML08p1RpYQ4_00043__id03854.mp4", "RVRA__real__153f__vocoded__id04278_ML08p1RpYQ4_00043__id03854.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A visible boundary line is present around the cheeks and jawline where the inner face meets the neck. Additionally, the skin tone and lighting on the inner face appear inconsistent with the surrounding forehead and neck from the start to the end of the clip.", "audio": "A slight metallic buzz is audible in the vocal track during speech. The overall voice timbre also creates a noticeable mismatch with the speaker on screen." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__animate__77f__vocoded__id00165_vWMMqjsL62A_00222__id05960", "clips": [ "FVRA__animate__77f__vocoded__id00165_vWMMqjsL62A_00222__id05960.mp4", "RVRA__real__77f__vocoded__id00165_vWMMqjsL62A_00222__id05960.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Unnatural blur appears around the outline of the blonde hair and shoulders against the dark background in the early frames. Minor texture jittering occurs along the hair edges and forehead during head movements toward the middle.", "audio": "The speech maintains a natural tone, smooth cadence, and clear pronunciation throughout the clip." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id06066__6z3Jqx3LnM_00134__id00657", "clips": [ "FVFA__animate__77f__main__id06066__6z3Jqx3LnM_00134__id00657.mp4", "RVRA__real__77f__vocoded__id06066__6z3Jqx3LnM_00134__id00657.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Near the start and middle of the clip, the edges around the headphones and hair display unnatural blur against the dark background. Near the middle as the head turns, the lower jawline and neck edge show slight shape distortion.", "audio": "From the start to the middle, the voice speaking into the microphone has a slight metallic timbre. Near the end of the phrase, the spoken English voice presents a clear voice-face mismatch with the French radio setup visible in the frame." }, "artifacts": { "video": [ "unnatural blur", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id03606_tqcv6AgbLWU_00265__id04900", "clips": [ "FVRA__animate__77f__vocoded__id03606_tqcv6AgbLWU_00265__id04900.mp4", "RVRA__real__77f__vocoded__id03606_tqcv6AgbLWU_00265__id04900.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the middle, the hair boundary fuzzes into the dark background on both sides. As the subject speaks near the center frames, the mouth contour deforms unnaturally.", "audio": "The vocal tone sounds smooth with natural intonation across the German utterance." }, "artifacts": { "video": [ "unnatural blur", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id05457_eGKOgBOXMpg_00082__id08352", "clips": [ "FVFA__animate__77f__main__id05457_eGKOgBOXMpg_00082__id08352.mp4", "RVRA__real__77f__vocoded__id05457_eGKOgBOXMpg_00082__id08352.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start to the end, the woman's hair and facial features show soft blurring and subtle texture jittering. Around the middle of the clip when her head turns slightly, minor shape distortion occurs along the jawline.", "audio": "Throughout the clip, the vocal timbre carries a buzzy quality on spoken words. The lower pitch also creates a noticeable voice-face mismatch with the person shown." }, "artifacts": { "video": [ "unnatural blur", "texture jittering", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__77f__main__id03615_c5yLzA-QV4s_00089__id05170", "clips": [ "FVFA__s2v__77f__main__id03615_c5yLzA-QV4s_00089__id05170.mp4", "RVRA__real__77f__vocoded__id03615_c5yLzA-QV4s_00089__id05170.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start to the middle, the head and mouth movements appear rigid against a completely static background. Towards the end, the lip shapes and teeth exhibit subtle warping and shape distortion as she speaks.", "audio": "Across the clip, the spoken vocal tone carries a faint metallic timbre during speech. The voice pitch and timbre also create a distinct voice-face mismatch with the presenter." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__facefusion__77f__main__id04495__k8mU1T87Pw_00127__id06471", "clips": [ "FVFA__facefusion__77f__main__id04495__k8mU1T87Pw_00127__id06471.mp4", "RVRA__real__77f__vocoded__id04495__k8mU1T87Pw_00127__id06471.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start to the end of the clip, a subtle line of altered skin texture and lighting is visible along the edge of the cheek and jawline. The inner facial skin appears smoother and less detailed than the surrounding neck and forehead.", "audio": "From the beginning to the end of the phrase, the voice carries a slightly metallic quality on sustained vowel sounds. The tone of the speech also displays a subtle mismatch with the person seen speaking." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id03681_Szvjxba6BTg_00064__id01664", "clips": [ "FVFA__s2v__153f__main__id03681_Szvjxba6BTg_00064__id01664.mp4", "RVRA__real__153f__vocoded__id03681_Szvjxba6BTg_00064__id01664.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The mouth motion appears rigid and poorly synced with speech in the middle frames. The background stays completely frozen while the head sways slightly. Facial skin textures display subtle jitter across frames.", "audio": "The speech timbre sounds unnatural with a buzzy quality on spoken words. The voice characteristics do not match the older speaker seen in the clip." }, "artifacts": { "video": [ "unnatural human movement", "texture jittering", "lip-sync mismatch" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__153f__main__id08803_SoxImR0IXkw_00054__id02359", "clips": [ "FVFA__facefusion__153f__main__id08803_SoxImR0IXkw_00054__id02359.mp4", "RVRA__real__153f__vocoded__id08803_SoxImR0IXkw_00054__id02359.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "A distinct skin tone discrepancy is visible along the inner cheek boundaries and hairline, creating an edge relative to the neck. The inner facial features also present a different identity from the surrounding head frame throughout the clip.", "audio": "A slight metallic quality affects the voice timbre during the middle spoken phrase on words like 'boyfriends' and 'discriminate'. The vocal acoustics also show a slight mismatch with the speaker shown in the frame." }, "artifacts": { "video": [ "blending seam", "person identity inconsistency" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__77f__main__id06429_IvbCTLC8B48_00061__id07574", "clips": [ "RVFA__seed_vc__77f__main__id06429_IvbCTLC8B48_00061__id07574.mp4", "RVRA__real__77f__vocoded__id06429_IvbCTLC8B48_00061__id07574.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "The lighting across the woman's profile and the movement of her lips remain natural and consistent from the start to the end.", "audio": "From the start to the middle, the spoken words exhibit a noticeably buzzy, metallic timbre. Throughout the phrase, the vocal texture creates a distinct mismatch with the woman visible on screen." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__153f__vocoded__id01643_7EiIGXK8VcM_00021__id06071", "clips": [ "RVRA__real__153f__vocoded__id01643_7EiIGXK8VcM_00021__id06071.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "In the middle frames, the speaker's mouth movements and subtle facial expressions transition naturally. The lighting across his glasses and forehead remains visually consistent from the early frames to the end.", "audio": "The speech maintains a natural tonal inflection and clear pronunciation throughout the clip, matching the speaker's visible mouth movement." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVFA__seed_vc__153f__main__id03854_19nf2F0wlIs_00001__id08581", "clips": [ "RVFA__seed_vc__153f__main__id03854_19nf2F0wlIs_00001__id08581.mp4", "RVRA__real__153f__vocoded__id03854_19nf2F0wlIs_00001__id08581.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Facial expressions, lip movements, and head motion remain natural and consistent throughout the clip. Skin textures and lighting look natural from start to end.", "audio": "The voice exhibits an unnatural metallic timbre on words like 'infectious' in the middle of the clip. Additionally, the vocal timbre does not naturally match the speaker seen in the video." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id06210_omDOOhIuYRA_00105__id07398", "clips": [ "FVRA__facefusion__77f__vocoded__id06210_omDOOhIuYRA_00105__id07398.mp4", "RVRA__real__77f__vocoded__id06210_omDOOhIuYRA_00105__id07398.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Along the cheek and jawline borders near the start, a slight color mismatch is visible between the inner face and neck. Around the middle of the clip, the skin texture around the mouth and chin shows unnatural blurring as the lips move.", "audio": "At the start of the clip, the voice exhibits natural pitch variations and tone. Toward the end, speech articulation remains clear and natural." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__77f__vocoded__id04597_jnl-F6GRZFc_00053__id03123", "clips": [ "FVRA__s2v__77f__vocoded__id04597_jnl-F6GRZFc_00053__id03123.mp4", "RVRA__real__77f__vocoded__id04597_jnl-F6GRZFc_00053__id03123.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, the background remains static and frozen despite slight head motions. Around the middle frames as the mouth opens wide, the chin and lower lip show noticeable shape distortion.", "audio": "The speaker's voice maintains a natural pitch contour and clear articulation throughout the spoken phrase." }, "artifacts": { "video": [ "shape distortion", "unnatural human movement" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__77f__vocoded__id03943_t5Aic_hjt0s_00159__id08046", "clips": [ "RVRA__real__77f__vocoded__id03943_t5Aic_hjt0s_00159__id08046.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "In the early frames, the speaker's lip movements align naturally with the spoken words as her mouth opens and closes. Throughout the movement, facial skin texture and lighting across the cheeks and forehead remain consistent.", "audio": "The vocal cadence and pitch flow naturally across the French utterance from start to end. Pronunciation of consonants and subtle breathing sounds remain clear and smooth." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id05170_rX_ilbVE9TY_00153__id05668", "clips": [ "FVRA__facefusion__153f__vocoded__id05170_rX_ilbVE9TY_00153__id05668.mp4", "RVRA__real__153f__vocoded__id05170_rX_ilbVE9TY_00153__id05668.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From frame 4 through the end, a dark patch with mismatched tone appears on the upper forehead near the hairline. When the mouth opens wide around frame 4, the inner facial skin displays unnatural blur and smoothing relative to the neck.", "audio": "The voice retains natural pitch contours and clean consonant articulation from the start to the end of the clip." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__s2v__153f__vocoded__id04070_4vfiyYW44jw_00033__id00657", "clips": [ "FVRA__s2v__153f__vocoded__id04070_4vfiyYW44jw_00033__id00657.mp4", "RVRA__real__153f__vocoded__id04070_4vfiyYW44jw_00033__id00657.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, the background behind the woman remains completely static despite her head motion. In the middle of the clip while speaking, her lip and facial movements display subtle shape distortion and artificial motion.", "audio": "The speech sounds clear and natural with consistent vocal tone and realistic phrasing from start to finish." }, "artifacts": { "video": [ "camera motion inconsistency", "unnatural human movement", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__animate__77f__vocoded__id04126_vQ9igU1IY2w_00232__id03710", "clips": [ "FVRA__animate__77f__vocoded__id04126_vQ9igU1IY2w_00232__id03710.mp4", "RVRA__real__77f__vocoded__id04126_vQ9igU1IY2w_00232__id03710.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "From the start to the end, subtle flickering and slight blurring appear along the outer edge of the hair and around the glasses frame.", "audio": "The audio features natural speech cadence, clear consonants, and consistent vocal pitch from start to end." }, "artifacts": { "video": [ "texture jittering", "unnatural blur" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id08170_Vh468tdQKlc_00022__id08320", "clips": [ "FVFA__animate__77f__main__id08170_Vh468tdQKlc_00022__id08320.mp4", "RVRA__real__77f__vocoded__id08170_Vh468tdQKlc_00022__id08320.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "When the hand enters at the bottom left around the middle frames, the fingers exhibit noticeable shape distortion. The edge of the short hair shows subtle texture jittering against the dark background as the head tilts near the end.", "audio": "The voice has a metallic timbre during the words 'he's funny' in the middle of the clip. The deep voice pitch also creates a distinct voice-face mismatch relative to the person shown." }, "artifacts": { "video": [ "shape distortion", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id06015_MynDvGyAQBw_00124__id04126", "clips": [ "FVFA__animate__77f__main__id06015_MynDvGyAQBw_00124__id04126.mp4", "RVRA__real__77f__vocoded__id06015_MynDvGyAQBw_00124__id04126.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start through the middle of the clip, facial features and glasses exhibit blurring and flickering texture around the moving mouth.", "audio": "At the start of the phrase, the spoken voice carries a slight buzzy, synthetic timbre during vowel transitions." }, "artifacts": { "video": [ "unnatural blur", "texture jittering" ], "audio": [ "metallic or buzzy timbre" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id08667_cmb4PTcnw7I_00053__id01296", "clips": [ "FVRA__animate__77f__vocoded__id08667_cmb4PTcnw7I_00053__id01296.mp4", "RVRA__real__77f__vocoded__id08667_cmb4PTcnw7I_00053__id01296.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The boundary around the blonde hair flickers against the background, and facial features exhibit mild texture jittering while speaking.", "audio": "The speech sounds natural with clear consonants and realistic pitch variation throughout." }, "artifacts": { "video": [ "texture jittering", "unnatural human movement" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__77f__main__id06068_WZ2Kqz_0SBQ_00052__id05960", "clips": [ "RVFA__seed_vc__77f__main__id06068_WZ2Kqz_0SBQ_00052__id05960.mp4", "RVRA__real__77f__vocoded__id06068_WZ2Kqz_0SBQ_00052__id05960.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From 00:00 to 00:03, the speaker's facial expressions, skin details, and ambient lighting remain completely natural as he speaks.", "audio": "From 00:00 to 00:03, the voice presents a clear voice-face mismatch with the speaker shown on screen. The audio additionally carries a slightly metallic or buzzy timbre on consonants." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__153f__vocoded__id04278_hMA4WwxSC_0_00055__id08581", "clips": [ "FVRA__s2v__153f__vocoded__id04278_hMA4WwxSC_0_00055__id08581.mp4", "RVRA__real__153f__vocoded__id04278_hMA4WwxSC_0_00055__id08581.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Throughout the clip, the facial expressions and lip movements exhibit stiff, robotic motion while the background stays completely frozen. Near the middle of the clip, the lower lip warps unnaturally as the speaker articulates words.", "audio": "The speech sounds natural with clear cadence, realistic pitch variations, and authentic vocal resonance from start to finish." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__153f__vocoded__id08590_UhRFZwnafIk_00070__id04236", "clips": [ "FVRA__facefusion__153f__vocoded__id08590_UhRFZwnafIk_00070__id04236.mp4", "RVRA__real__153f__vocoded__id08590_UhRFZwnafIk_00070__id04236.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the middle of the clip, a faint edge and subtle skin tone shift are noticeable along the cheek and jawline where the inner face joins the surrounding face and neck.", "audio": "The spoken speech retains a natural vocal tone, consistent pitch, and clear sound quality from start to end." }, "artifacts": { "video": [ "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__s2v__77f__main__id04278_xKF4BM2trRY_00078__id00948", "clips": [ "FVFA__s2v__77f__main__id04278_xKF4BM2trRY_00078__id00948.mp4", "RVRA__real__77f__vocoded__id04278_xKF4BM2trRY_00078__id00948.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Near the middle of the clip, a hand briefly appears in the lower left frame with unnatural motion. The speaker's mouth exhibits subtle shape distortion around the lips while speaking.", "audio": "The speech exhibits a slightly unnatural, metallic timbre throughout the utterance. The vocal characteristics also display a noticeable voice-face mismatch relative to the visual speaker." }, "artifacts": { "video": [ "unnatural human movement", "shape distortion" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__s2v__153f__main__id09218_Tc4w_QRu6Ss_00148__id03084", "clips": [ "FVFA__s2v__153f__main__id09218_Tc4w_QRu6Ss_00148__id03084.mp4", "RVRA__real__153f__vocoded__id09218_Tc4w_QRu6Ss_00148__id03084.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_s2v", "regions": [ "whole frame", "voice" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "As the subject speaks from the start to the middle, the blue backdrop stays artificially motionless relative to head movements. Near the middle frames, the lip and mouth movements show subtle shape distortion and texture jittering.", "audio": "The voice exhibits a noticeable metallic or buzzy timbre throughout the speech. The tone also produces a voice-face mismatch relative to the man in the video." }, "artifacts": { "video": [ "camera motion inconsistency", "shape distortion", "texture jittering" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVRA__facefusion__77f__vocoded__id02176_vGVru0wHaYQ_00126__id08170", "clips": [ "FVRA__facefusion__77f__vocoded__id02176_vGVru0wHaYQ_00126__id08170.mp4", "RVRA__real__77f__vocoded__id02176_vGVru0wHaYQ_00126__id08170.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_facefusion", "regions": [ "inner face" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Along the cheekbones and jawline, the inner face displays a noticeable skin tone and lighting mismatch relative to the neck. A faint color boundary is visible where the face meets the original hairline and chin throughout the clip.", "audio": "The speaker's voice maintains a natural cadence, consistent pitch, and clear sound quality from the start to the end of the clip." }, "artifacts": { "video": [ "blending seam", "lighting inconsistency" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVRA__real__77f__vocoded__id01156_ULRf2FN7_w4_00118__id07430", "clips": [ "RVRA__real__77f__vocoded__id01156_ULRf2FN7_w4_00118__id07430.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Facial expressions and lip movements around the mouth align naturally as the head tilts at the start. Lighting across the face and background remains consistent near the middle.", "audio": "Vocal tone and pitch inflections sound natural during the opening words. Speech cadence and breath sounds remain clear near the end." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_RVFA__seed_vc__153f__main__id07780_PHB2HY7wghs_00015__id00144", "clips": [ "RVFA__seed_vc__153f__main__id07780_PHB2HY7wghs_00015__id00144.mp4", "RVRA__real__153f__vocoded__id07780_PHB2HY7wghs_00015__id00144.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVFA", "cell": "RVFA_real", "regions": [ "voice" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From start to end, facial expressions and lip movements remain natural and consistent with the lighting across her face.", "audio": "Between 0:01 and 0:05, the voice exhibits a buzzy, metallic timbre that creates a noticeable voice-face mismatch with the speaker." }, "artifacts": { "video": [], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__153f__main__id01156_q7X4wSOFcEY_00179__id01104", "clips": [ "FVFA__facefusion__153f__main__id01156_q7X4wSOFcEY_00179__id01104.mp4", "RVRA__real__153f__vocoded__id01156_q7X4wSOFcEY_00179__id01104.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "Around the middle of the clip, a subtle texture difference appears along the lower jawline and cheek where the face boundary meets the neck. When the subject speaks, the inner facial skin appears noticeably smoother than the surrounding skin.", "audio": "The speaker's voice exhibits a slight metallic timbre on the spoken words near the beginning and middle of the clip. The voice profile sounds acoustically detached from the person shown on screen." }, "artifacts": { "video": [ "blending seam", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 153 }, { "id": "t4_RVRA__real__153f__vocoded__id06480_YQD4AsloKcY_00078__id01534", "clips": [ "RVRA__real__153f__vocoded__id06480_YQD4AsloKcY_00078__id01534.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "RVRA", "cell": "RVRA_real", "regions": [ "none" ], "facts": { "video": { "fake": false, "region": "none", "changed": null, "original": "a camera recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "The speaker's head and lip movements align naturally with his speech while talking into the microphone near the start and middle frames. Facial lighting and skin textures on the cheek and jaw remain consistent as his expression changes.", "audio": "The spoken cadence and pitch contours sound natural and smooth, with clear consonant pronunciation and natural pauses between phrases." }, "artifacts": { "video": [], "audio": [] }, "no_evidence": [], "frames": 153 }, { "id": "t4_FVFA__facefusion__77f__main__id05877_6QUcq_6B-LU_00011__id07768", "clips": [ "FVFA__facefusion__77f__main__id05877_6QUcq_6B-LU_00011__id07768.mp4", "RVRA__real__77f__vocoded__id05877_6QUcq_6B-LU_00011__id07768.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_facefusion", "regions": [ "inner face", "voice" ], "facts": { "video": { "fake": true, "region": "inner face", "changed": "only the inner face was swapped with another person's face", "original": "hair, head shape, body, background and all movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "From the start through the middle of the clip, a dark mustache and inner face features appear superimposed with visible edge boundaries along the upper lip and cheeks. When the speaker turns his head near the end, the inner face skin texture and tone differ noticeably from the surrounding jaw and neck.", "audio": "Across the spoken phrase, the voice exhibits a distinctly buzzy, metallic timbre during vowel sounds in the middle of the clip. The resulting voice quality creates an unnatural audio match for the speaker shown on screen." }, "artifacts": { "video": [ "blending seam", "person identity inconsistency" ], "audio": [ "metallic or buzzy timbre", "voice-face mismatch" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__s2v__77f__vocoded__id04528_UF0U3evnpC4_00065__id03281", "clips": [ "FVRA__s2v__77f__vocoded__id04528_UF0U3evnpC4_00065__id03281.mp4", "RVRA__real__77f__vocoded__id04528_UF0U3evnpC4_00065__id03281.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_s2v", "regions": [ "whole frame" ], "facts": { "video": { "fake": true, "region": "whole frame", "changed": "the whole frame was generated from one photo, person and background", "original": "nothing in the picture is from a recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "At the start, the head and hair exhibit unnaturally rigid movement. Around the middle of the clip, subtle texture warping appears along the edges of the hair against the background as the mouth speaks.", "audio": "The speaker's voice maintains natural pitch transitions and clear consonant sound throughout the clip." }, "artifacts": { "video": [ "unnatural human movement", "texture jittering" ], "audio": [] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVFA__animate__77f__main__id03084_X1W-wTD6O2M_00038__id09218", "clips": [ "FVFA__animate__77f__main__id03084_X1W-wTD6O2M_00038__id09218.mp4", "RVRA__real__77f__vocoded__id03084_X1W-wTD6O2M_00038__id09218.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVFA", "cell": "FVFA_animate", "regions": [ "whole person", "voice" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": true, "region": "voice", "changed": "the voice timbre was converted to another person's voice (Seed-VC)", "original": "the words, timing and pauses are from the original recording" } }, "evidence": { "video": "At the start and middle of the clip, subtle texture jittering appears across the cheeks. Towards the end, the outer edge of the head exhibits soft, unnatural blur against the background wall.", "audio": "At the start while uttering '10 years old', the speech tone carries a slightly buzzy timbre. Near the end on the word 'acting', the voice exhibits an unnatural pitch shift." }, "artifacts": { "video": [ "texture jittering", "unnatural blur" ], "audio": [ "metallic or buzzy timbre", "unnatural pitch or loudness" ] }, "no_evidence": [], "frames": 77 }, { "id": "t4_FVRA__animate__77f__vocoded__id05538_Thg55gSVGIo_00092__id06536", "clips": [ "FVRA__animate__77f__vocoded__id05538_Thg55gSVGIo_00092__id06536.mp4", "RVRA__real__77f__vocoded__id05538_Thg55gSVGIo_00092__id06536.mp4" ], "question": "Which parts of this audio-visual clip, if any, were generated or altered by AI, where, and what in the clip shows it? Normal processing such as compression does not count.", "options": [ "none", "inner face", "whole person", "whole frame", "voice" ], "label": "FVRA", "cell": "FVRA_animate", "regions": [ "whole person" ], "facts": { "video": { "fake": true, "region": "whole person", "changed": "the whole person was regenerated as another person", "original": "the background, lighting and camera movement are from the original recording" }, "audio": { "fake": false, "region": "none", "changed": null, "original": "the speaker's real voice, passed through a vocoder only" } }, "evidence": { "video": "Around the middle frames, subtle texture flickering appears along the glasses frames and jawline against the backdrop. Near the end of the clip, soft edge blurring is visible along the hair boundaries and shirt collar during head movement.", "audio": "The speaker's voice maintains natural pitch transitions and smooth vocal resonance throughout the phrase. Consonant sounds and cadence remain clear and natural as he pronounces 'beans and rice'." }, "artifacts": { "video": [ "texture jittering", "unnatural blur", "blending seam" ], "audio": [] }, "no_evidence": [], "frames": 77 } ] } } }