[{"id": "t4_FVFA__animate__77f__main__id05427_bGkE8GV1Tbo_00173__id07525", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/animate/77f/main/FVFA__animate__77f__main__id05427_bGkE8GV1Tbo_00173__id07525.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real person's face would. The voice has been altered by AI.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.703}, {"artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording", "score": "melt", "value": 0.646}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real person's face would.", "audio": "The voice has been altered by AI."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording"}]}, "meta": {"video_method": "animate", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id05427", "target_id": "id07525", "pair_id": "id05427_bGkE8GV1Tbo_00173__id07525", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__animate__77f__main__id05427_bGkE8GV1Tbo_00173__id07525.mp4"}, {"id": "t4_FVRA__animate__77f__vocoded__id05170_0q6kg61DpZ0_00008__id03615", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/animate/77f/vocoded/FVRA__animate__77f__vocoded__id05170_0q6kg61DpZ0_00008__id03615.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, covering both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to. The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": []}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, covering both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed."}, "evidence": []}, "meta": {"video_method": "animate", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id05170", "target_id": "id03615", "pair_id": "id05170_0q6kg61DpZ0_00008__id03615", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__animate__77f__vocoded__id05170_0q6kg61DpZ0_00008__id03615.mp4"}, {"id": "t4_FVRA__animate__77f__vocoded__id04278_hMA4WwxSC_0_00056__id08581", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/animate/77f/vocoded/FVRA__animate__77f__vocoded__id04278_hMA4WwxSC_0_00056__id08581.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail. The audio sounds like a natural, unaltered recording, with nothing in it that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.73}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in it that has been changed."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "animate", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id04278", "target_id": "id08581", "pair_id": "id04278_hMA4WwxSC_0_00056__id08581", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__animate__77f__vocoded__id04278_hMA4WwxSC_0_00056__id08581.mp4"}, {"id": "t4_FVFA__animate__77f__main__id08406_rlGlwmTEi0I_00235__id08808", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/animate/77f/main/FVFA__animate__77f__main__id08406_rlGlwmTEi0I_00235__id08808.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The brightness visibly jumps from one frame to the next, so the lighting flickers instead of staying steady. The voice is AI-altered. Nothing specific stands out as wrong when listening.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.543}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The brightness visibly jumps from one frame to the next, so the lighting flickers instead of staying steady.", "audio": "The voice is AI-altered. Nothing specific stands out as wrong when listening."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id08406", "target_id": "id08808", "pair_id": "id08406_rlGlwmTEi0I_00235__id08808", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__animate__77f__main__id08406_rlGlwmTEi0I_00235__id08808.mp4"}, {"id": "t4_RVRA__real__77f__vocoded__id01348_XyfOKBixCpw_00037__id09188", "task": "task4_explanation", "split": "test", "clips": ["videos/RVRA/real/77f/vocoded/RVRA__real__77f__vocoded__id01348_XyfOKBixCpw_00037__id09188.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording. Movement, lighting and detail stay consistent from start to finish, and nothing in the picture appears generated or changed. The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and consistent, with nothing that seems generated or changed.", "gt_label": "RVRA", "gt_target": {"fake_parts": [], "regions": ["none"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The video looks like a natural, unaltered recording. Movement, lighting and detail stay consistent from start to finish, and nothing in the picture appears generated or changed.", "audio": "The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and consistent, with nothing that seems generated or changed."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id01348", "target_id": "id09188", "pair_id": "id01348_XyfOKBixCpw_00037__id09188", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVRA__real__77f__vocoded__id01348_XyfOKBixCpw_00037__id09188.mp4"}, {"id": "t4_RVRA__real__77f__vocoded__id04717_7DiEgpK7dbk_00019__id00761", "task": "task4_explanation", "split": "test", "clips": ["videos/RVRA/real/77f/vocoded/RVRA__real__77f__vocoded__id04717_7DiEgpK7dbk_00019__id00761.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording. Movement, lighting and detail all hold together from start to finish, and nothing in the picture appears generated or edited. The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and organic, with nothing that suggests it was generated or changed.", "gt_label": "RVRA", "gt_target": {"fake_parts": [], "regions": ["none"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The video looks like a natural, unaltered recording. Movement, lighting and detail all hold together from start to finish, and nothing in the picture appears generated or edited.", "audio": "The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and organic, with nothing that suggests it was generated or changed."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id04717", "target_id": "id00761", "pair_id": "id04717_7DiEgpK7dbk_00019__id00761", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVRA__real__77f__vocoded__id04717_7DiEgpK7dbk_00019__id00761.mp4"}, {"id": "t4_FVFA__s2v__77f__main__id08808_pIEXYlx5_XE_00024__id07630", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/s2v/77f/main/FVFA__s2v__77f__main__id08808_pIEXYlx5_XE_00024__id07630.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft, smeared detail. It also wobbles and warps as the person moves, shifting shape more than a real face would. The voice has been AI-altered. No specific sign in the sound is given beyond that.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.671}, {"artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording", "score": "melt", "value": 1.292}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft, smeared detail. It also wobbles and warps as the person moves, shifting shape more than a real face would.", "audio": "The voice has been AI-altered. No specific sign in the sound is given beyond that."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id08808", "target_id": "id07630", "pair_id": "id08808_pIEXYlx5_XE_00024__id07630", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__s2v__77f__main__id08808_pIEXYlx5_XE_00024__id07630.mp4"}, {"id": "t4_FVRA__facefusion__77f__vocoded__id01664_ETpbNS7U3gY_00111__id03681", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/facefusion/77f/vocoded/FVRA__facefusion__77f__vocoded__id01664_ETpbNS7U3gY_00111__id03681.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so its details come across soft and indistinct. The audio sounds like a natural, unaltered recording, with nothing in it that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["inner face"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.806}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so its details come across soft and indistinct.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in it that has been changed."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id01664", "target_id": "id03681", "pair_id": "id01664_ETpbNS7U3gY_00111__id03681", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__facefusion__77f__vocoded__id01664_ETpbNS7U3gY_00111__id03681.mp4"}, {"id": "t4_FVFA__facefusion__77f__main__id08315_WdCsuUtH5lE_00054__id05170", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/facefusion/77f/main/FVFA__facefusion__77f__main__id08315_WdCsuUtH5lE_00054__id05170.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with soft, smudged detail. The voice is AI-altered. No specific sound artifact stands out.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["inner face", "voice"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.75}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with soft, smudged detail.", "audio": "The voice is AI-altered. No specific sound artifact stands out."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id08315", "target_id": "id05170", "pair_id": "id08315_WdCsuUtH5lE_00054__id05170", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__facefusion__77f__main__id08315_WdCsuUtH5lE_00054__id05170.mp4"}, {"id": "t4_FVRA__s2v__77f__vocoded__id08610_bNkwm5lMs3Q_00057__id01052", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/s2v/77f/vocoded/FVRA__s2v__77f__vocoded__id08610_bNkwm5lMs3Q_00057__id01052.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to. The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": []}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed."}, "evidence": []}, "meta": {"video_method": "s2v", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id08610", "target_id": "id01052", "pair_id": "id08610_bNkwm5lMs3Q_00057__id01052", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__s2v__77f__vocoded__id08610_bNkwm5lMs3Q_00057__id01052.mp4"}, {"id": "t4_RVFA__seed_vc__77f__main__id02255_W8B7APZ3Oo8_00026__id02472", "task": "task4_explanation", "split": "test", "clips": ["videos/RVFA/seed_vc/77f/main/RVFA__seed_vc__77f__main__id02255_W8B7APZ3Oo8_00026__id02472.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording, with nothing in the picture that has been changed. The voice is AI-generated, while the video itself is untouched. No specific audible flaws stand out in it.", "gt_label": "RVFA", "gt_target": {"fake_parts": ["audio"], "regions": ["voice"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The video looks like a natural, unaltered recording, with nothing in the picture that has been changed.", "audio": "The voice is AI-generated, while the video itself is untouched. No specific audible flaws stand out in it."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id02255", "target_id": "id02472", "pair_id": "id02255_W8B7APZ3Oo8_00026__id02472", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVFA__seed_vc__77f__main__id02255_W8B7APZ3Oo8_00026__id02472.mp4"}, {"id": "t4_FVRA__s2v__77f__vocoded__id00144_myrnp6zXitU_00151__id05042", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/s2v/77f/vocoded/FVRA__s2v__77f__vocoded__id00144_myrnp6zXitU_00151__id05042.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps as the clip plays, shifting its shape more than a face in a real recording would. The audio sounds like a natural, unaltered recording, with nothing in it changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording", "score": "melt", "value": 1.366}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps as the clip plays, shifting its shape more than a face in a real recording would.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in it changed."}, "evidence": [{"modality": "video", "artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id00144", "target_id": "id05042", "pair_id": "id00144_myrnp6zXitU_00151__id05042", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__s2v__77f__vocoded__id00144_myrnp6zXitU_00151__id05042.mp4"}, {"id": "t4_RVFA__seed_vc__77f__main__id03613_y-weegLHBmA_00146__id00476", "task": "task4_explanation", "split": "test", "clips": ["videos/RVFA/seed_vc/77f/main/RVFA__seed_vc__77f__main__id03613_y-weegLHBmA_00146__id00476.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording, and nothing in the picture appears generated or changed. The voice is AI-altered, while the footage itself is untouched. No specific audible flaw stands out in it.", "gt_label": "RVFA", "gt_target": {"fake_parts": ["audio"], "regions": ["voice"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The video looks like a natural, unaltered recording, and nothing in the picture appears generated or changed.", "audio": "The voice is AI-altered, while the footage itself is untouched. No specific audible flaw stands out in it."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id03613", "target_id": "id00476", "pair_id": "id03613_y-weegLHBmA_00146__id00476", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVFA__seed_vc__77f__main__id03613_y-weegLHBmA_00146__id00476.mp4"}, {"id": "t4_FVRA__facefusion__77f__vocoded__id05096_mrhzD6Y5OnA_00098__id00330", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/facefusion/77f/vocoded/FVRA__facefusion__77f__vocoded__id05096_mrhzD6Y5OnA_00098__id00330.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, and its brightness jumps from one frame to the next. The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["inner face"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.797}, {"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.401}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, and its brightness jumps from one frame to the next.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "facefusion", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id05096", "target_id": "id00330", "pair_id": "id05096_mrhzD6Y5OnA_00098__id00330", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__facefusion__77f__vocoded__id05096_mrhzD6Y5OnA_00098__id00330.mp4"}, {"id": "t4_FVFA__s2v__77f__main__id00761_C_ZBK3E8bJk_00027__id07616", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/s2v/77f/main/FVFA__s2v__77f__main__id00761_C_ZBK3E8bJk_00027__id07616.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps as the clip plays, shifting its shape more than a face in a real recording would. The voice is AI-altered. No specific sign in the sound is given beyond that.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording", "score": "melt", "value": 1.255}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps as the clip plays, shifting its shape more than a face in a real recording would.", "audio": "The voice is AI-altered. No specific sign in the sound is given beyond that."}, "evidence": [{"modality": "video", "artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id00761", "target_id": "id07616", "pair_id": "id00761_C_ZBK3E8bJk_00027__id07616", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__s2v__77f__main__id00761_C_ZBK3E8bJk_00027__id07616.mp4"}, {"id": "t4_FVFA__facefusion__77f__main__id06210_ZG5rWainMdg_00072__id01643", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/facefusion/77f/main/FVFA__facefusion__77f__main__id06210_ZG5rWainMdg_00072__id01643.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so fine detail in the features and skin is soft. The voice is AI-altered. No specific sign in the sound is given beyond that.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["inner face", "voice"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.802}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so fine detail in the features and skin is soft.", "audio": "The voice is AI-altered. No specific sign in the sound is given beyond that."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id06210", "target_id": "id01643", "pair_id": "id06210_ZG5rWainMdg_00072__id01643", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__facefusion__77f__main__id06210_ZG5rWainMdg_00072__id01643.mp4"}, {"id": "t4_FVFA__s2v__153f__main__id06071_fZTqMIYecnk_00163__id06210", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/s2v/153f/main/FVFA__s2v__153f__main__id06071_fZTqMIYecnk_00163__id06210.mov"], "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 frame", "voice"], "gt_answer": "The whole frame, both the person and the background, is AI-generated. The brightness visibly jumps from one frame to the next, so the lighting flickers instead of staying steady. The voice has been altered by AI. Nothing else is given about what sounds wrong in it.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.452}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame, both the person and the background, is AI-generated. The brightness visibly jumps from one frame to the next, so the lighting flickers instead of staying steady.", "audio": "The voice has been altered by AI. Nothing else is given about what sounds wrong in it."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "s2v", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id06071", "target_id": "id06210", "pair_id": "id06071_fZTqMIYecnk_00163__id06210", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__s2v__153f__main__id06071_fZTqMIYecnk_00163__id06210.mp4"}, {"id": "t4_RVRA__real__153f__vocoded__id00509_tpDTuxBwXBs_00065__id04278", "task": "task4_explanation", "split": "test", "clips": ["videos/RVRA/real/153f/vocoded/RVRA__real__153f__vocoded__id00509_tpDTuxBwXBs_00065__id04278.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording. Movement, lighting and detail stay consistent from start to finish, and nothing in the picture appears generated or changed. The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and consistent, with nothing that seems generated or edited.", "gt_label": "RVRA", "gt_target": {"fake_parts": [], "regions": ["none"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The video looks like a natural, unaltered recording. Movement, lighting and detail stay consistent from start to finish, and nothing in the picture appears generated or changed.", "audio": "The audio sounds like a natural, unaltered recording. The voice and any background sound come across as continuous and consistent, with nothing that seems generated or edited."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id00509", "target_id": "id04278", "pair_id": "id00509_tpDTuxBwXBs_00065__id04278", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVRA__real__153f__vocoded__id00509_tpDTuxBwXBs_00065__id04278.mp4"}, {"id": "t4_FVFA__facefusion__153f__main__id06471__ZRFVCmOwNM_00177__id04495", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/facefusion/153f/main/FVFA__facefusion__153f__main__id06471__ZRFVCmOwNM_00177__id04495.mov"], "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 frame", "voice"], "gt_answer": "The inner face (eyes, nose, mouth and skin) is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with soft, smudged detail. The voice is AI-altered. No specific sign was measured for it, so nothing here can be pointed to beyond the voice itself being generated.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["inner face", "voice"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.758}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The inner face (eyes, nose, mouth and skin) is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with soft, smudged detail.", "audio": "The voice is AI-altered. No specific sign was measured for it, so nothing here can be pointed to beyond the voice itself being generated."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id06471", "target_id": "id04495", "pair_id": "id06471__ZRFVCmOwNM_00177__id04495", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__facefusion__153f__main__id06471__ZRFVCmOwNM_00177__id04495.mp4"}, {"id": "t4_FVRA__s2v__153f__vocoded__id07054_uFcEE5wJrOU_00063__id04236", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/s2v/153f/vocoded/FVRA__s2v__153f__vocoded__id07054_uFcEE5wJrOU_00063__id04236.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face moves noticeably less than it would in a real recording, so its expressions and motion look restrained and somewhat stiff. The audio sounds like a natural recording of a real voice, with nothing altered.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording", "score": "melt", "value": 0.578}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face moves noticeably less than it would in a real recording, so its expressions and motion look restrained and somewhat stiff.", "audio": "The audio sounds like a natural recording of a real voice, with nothing altered."}, "evidence": [{"modality": "video", "artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id07054", "target_id": "id04236", "pair_id": "id07054_uFcEE5wJrOU_00063__id04236", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__s2v__153f__vocoded__id07054_uFcEE5wJrOU_00063__id04236.mp4"}, {"id": "t4_RVRA__real__153f__vocoded__id02472_GZgukKKkItg_00032__id01708", "task": "task4_explanation", "split": "test", "clips": ["videos/RVRA/real/153f/vocoded/RVRA__real__153f__vocoded__id02472_GZgukKKkItg_00032__id01708.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural recording, with nothing altered. The motion, lighting and detail all hold together from start to finish, and nothing in it looks generated or edited. The audio sounds like a natural recording, with nothing altered. The voice and any background sound are steady and sound like they were captured live.", "gt_label": "RVRA", "gt_target": {"fake_parts": [], "regions": ["none"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The video looks like a natural recording, with nothing altered. The motion, lighting and detail all hold together from start to finish, and nothing in it looks generated or edited.", "audio": "The audio sounds like a natural recording, with nothing altered. The voice and any background sound are steady and sound like they were captured live."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id02472", "target_id": "id01708", "pair_id": "id02472_GZgukKKkItg_00032__id01708", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVRA__real__153f__vocoded__id02472_GZgukKKkItg_00032__id01708.mp4"}, {"id": "t4_FVRA__facefusion__153f__vocoded__id00948_Qppg8ZCTQLE_00038__id00509", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/facefusion/153f/vocoded/FVRA__facefusion__153f__vocoded__id00948_Qppg8ZCTQLE_00038__id00509.mov"], "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 frame", "voice"], "gt_answer": "The eyes, nose, mouth and skin of the face are AI-generated, while the hair, head shape, body and background are untouched. The face's brightness jumps from one frame to the next, flickering lighter and darker in a way that breaks the steady look of the footage. The voice sounds like a natural recording, with nothing altered.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["inner face"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.439}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The eyes, nose, mouth and skin of the face are AI-generated, while the hair, head shape, body and background are untouched. The face's brightness jumps from one frame to the next, flickering lighter and darker in a way that breaks the steady look of the footage.", "audio": "The voice sounds like a natural recording, with nothing altered."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "facefusion", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id00948", "target_id": "id00509", "pair_id": "id00948_Qppg8ZCTQLE_00038__id00509", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__facefusion__153f__vocoded__id00948_Qppg8ZCTQLE_00038__id00509.mp4"}, {"id": "t4_FVFA__facefusion__153f__main__id07839_41lss_yYWIM_00008__id05768", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/facefusion/153f/main/FVFA__facefusion__153f__main__id07839_41lss_yYWIM_00008__id05768.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so its detail is soft. The voice has been altered by AI. No specific sound artifacts are noticeable in it.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["inner face", "voice"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.824}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, so its detail is soft.", "audio": "The voice has been altered by AI. No specific sound artifacts are noticeable in it."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id07839", "target_id": "id05768", "pair_id": "id07839_41lss_yYWIM_00008__id05768", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__facefusion__153f__main__id07839_41lss_yYWIM_00008__id05768.mp4"}, {"id": "t4_FVRA__facefusion__153f__vocoded__id06713_5qTLKWiq73k_00072__id05042", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/facefusion/153f/vocoded/FVRA__facefusion__153f__vocoded__id06713_5qTLKWiq73k_00072__id05042.mov"], "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 frame", "voice"], "gt_answer": "The eyes, nose, mouth and skin of the face are AI-generated, while the hair, head shape, body and background are untouched. The brightness of the face jumps from one frame to the next, so it flickers lighter and darker instead of holding steady. The audio sounds like a natural recording, with nothing altered.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["inner face"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.755}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The eyes, nose, mouth and skin of the face are AI-generated, while the hair, head shape, body and background are untouched. The brightness of the face jumps from one frame to the next, so it flickers lighter and darker instead of holding steady.", "audio": "The audio sounds like a natural recording, with nothing altered."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "facefusion", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id06713", "target_id": "id05042", "pair_id": "id06713_5qTLKWiq73k_00072__id05042", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__facefusion__153f__vocoded__id06713_5qTLKWiq73k_00072__id05042.mp4"}, {"id": "t4_FVFA__s2v__153f__main__id01568_h7aR8x6JtDk_00081__id08315", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/s2v/153f/main/FVFA__s2v__153f__main__id01568_h7aR8x6JtDk_00081__id08315.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, including the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail. The voice has been altered by AI. Nothing else about the audio is described beyond the voice itself being synthetic.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.673}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, including the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail.", "audio": "The voice has been altered by AI. Nothing else about the audio is described beyond the voice itself being synthetic."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id01568", "target_id": "id08315", "pair_id": "id01568_h7aR8x6JtDk_00081__id08315", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__s2v__153f__main__id01568_h7aR8x6JtDk_00081__id08315.mp4"}, {"id": "t4_FVRA__s2v__153f__vocoded__id08046_9dBnzvP4A9A_00022__id03943", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/s2v/153f/vocoded/FVRA__s2v__153f__vocoded__id08046_9dBnzvP4A9A_00022__id03943.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, covering both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to. The audio sounds like a natural, unaltered recording, with nothing in the voice or the sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": []}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, covering both the person and the background. Nothing in it looks visibly wrong, so no specific flaw can be pointed to.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or the sound that has been changed."}, "evidence": []}, "meta": {"video_method": "s2v", "audio_method": "bigvgan", "frames": 153, "arm": "vocoded", "source_id": "id08046", "target_id": "id03943", "pair_id": "id08046_9dBnzvP4A9A_00022__id03943", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__s2v__153f__vocoded__id08046_9dBnzvP4A9A_00022__id03943.mp4"}, {"id": "t4_RVFA__seed_vc__153f__main__id02255_DA5DbluCDOs_00008__id06758", "task": "task4_explanation", "split": "test", "clips": ["videos/RVFA/seed_vc/153f/main/RVFA__seed_vc__153f__main__id02255_DA5DbluCDOs_00008__id06758.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording, with nothing in the picture changed or generated. The voice is AI-generated; the speech itself has been altered by AI. The video is untouched, so only the audio carries the manipulation.", "gt_label": "RVFA", "gt_target": {"fake_parts": ["audio"], "regions": ["voice"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The video looks like a natural, unaltered recording, with nothing in the picture changed or generated.", "audio": "The voice is AI-generated; the speech itself has been altered by AI. The video is untouched, so only the audio carries the manipulation."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id02255", "target_id": "id06758", "pair_id": "id02255_DA5DbluCDOs_00008__id06758", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVFA__seed_vc__153f__main__id02255_DA5DbluCDOs_00008__id06758.mp4"}, {"id": "t4_RVFA__seed_vc__153f__main__id06015_huOOE9WL2pk_00209__id01376", "task": "task4_explanation", "split": "test", "clips": ["videos/RVFA/seed_vc/153f/main/RVFA__seed_vc__153f__main__id06015_huOOE9WL2pk_00209__id01376.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording, with nothing in the picture appearing generated or changed. The voice is AI-generated; the speech itself has been altered, while the footage it plays over is untouched. No specific oddities in the sound stand out beyond that.", "gt_label": "RVFA", "gt_target": {"fake_parts": ["audio"], "regions": ["voice"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The video looks like a natural, unaltered recording, with nothing in the picture appearing generated or changed.", "audio": "The voice is AI-generated; the speech itself has been altered, while the footage it plays over is untouched. No specific oddities in the sound stand out beyond that."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "seed_vc", "frames": 153, "arm": "main", "source_id": "id06015", "target_id": "id01376", "pair_id": "id06015_huOOE9WL2pk_00209__id01376", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVFA__seed_vc__153f__main__id06015_huOOE9WL2pk_00209__id01376.mp4"}, {"id": "t4_FVFA__animate__77f__main__id01505_dVKZSQDG9X4_00075__id00762", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/animate/77f/main/FVFA__animate__77f__main__id01505_dVKZSQDG9X4_00075__id00762.mov"], "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 frame", "voice"], "gt_answer": "The entire frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real face normally would. The brightness also jumps noticeably from one frame to the next. The voice has been altered by AI.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.687}, {"artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording", "score": "melt", "value": 0.777}, {"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.658}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The entire frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real face normally would. The brightness also jumps noticeably from one frame to the next.", "audio": "The voice has been altered by AI."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording"}, {"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id01505", "target_id": "id00762", "pair_id": "id01505_dVKZSQDG9X4_00075__id00762", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__animate__77f__main__id01505_dVKZSQDG9X4_00075__id00762.mp4"}, {"id": "t4_FVRA__animate__77f__vocoded__id01534_zaac3OQqEQE_00097__id06480", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/animate/77f/vocoded/FVRA__animate__77f__vocoded__id01534_zaac3OQqEQE_00097__id06480.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, covering both the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail. The audio sounds like a natural, unaltered recording, with nothing in it that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.557}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, covering both the person and the background. The face looks blurrier than it would in a real recording, with soft, indistinct detail.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in it that has been changed."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "animate", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id01534", "target_id": "id06480", "pair_id": "id01534_zaac3OQqEQE_00097__id06480", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__animate__77f__vocoded__id01534_zaac3OQqEQE_00097__id06480.mp4"}, {"id": "t4_FVFA__animate__77f__main__id03943_t5Aic_hjt0s_00155__id08046", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/animate/77f/main/FVFA__animate__77f__main__id03943_t5Aic_hjt0s_00155__id08046.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real face normally would. The brightness also jumps noticeably from one frame to the next. The voice has been altered by AI. Nothing else is given about how it sounds, so this is the only part that is not a natural recording.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.556}, {"artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording", "score": "melt", "value": 0.77}, {"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.536}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and it moves less than a real face normally would. The brightness also jumps noticeably from one frame to the next.", "audio": "The voice has been altered by AI. Nothing else is given about how it sounds, so this is the only part that is not a natural recording."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "unnatural human movement", "meaning": "the face moves less than in a real recording"}, {"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id03943", "target_id": "id08046", "pair_id": "id03943_t5Aic_hjt0s_00155__id08046", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__animate__77f__main__id03943_t5Aic_hjt0s_00155__id08046.mp4"}, {"id": "t4_FVRA__animate__77f__vocoded__id03123_3GdossNwwVs_00011__id05623", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/animate/77f/vocoded/FVRA__animate__77f__vocoded__id03123_3GdossNwwVs_00011__id05623.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The brightness jumps noticeably from one frame to the next, so the lighting flickers instead of staying steady. The audio sounds like a natural recording, with nothing altered.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.753}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The brightness jumps noticeably from one frame to the next, so the lighting flickers instead of staying steady.", "audio": "The audio sounds like a natural recording, with nothing altered."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id03123", "target_id": "id05623", "pair_id": "id03123_3GdossNwwVs_00011__id05623", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__animate__77f__vocoded__id03123_3GdossNwwVs_00011__id05623.mp4"}, {"id": "t4_FVFA__animate__77f__main__id03606_tqcv6AgbLWU_00265__id04900", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/animate/77f/main/FVFA__animate__77f__main__id03606_tqcv6AgbLWU_00265__id04900.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and the brightness jumps noticeably from one frame to the next. The voice has been altered by AI.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.695}, {"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.51}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, and the brightness jumps noticeably from one frame to the next.", "audio": "The voice has been altered by AI."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id03606", "target_id": "id04900", "pair_id": "id03606_tqcv6AgbLWU_00265__id04900", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__animate__77f__main__id03606_tqcv6AgbLWU_00265__id04900.mp4"}, {"id": "t4_FVRA__animate__77f__vocoded__id02359_QiCI0mefCx4_00042__id03078", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/animate/77f/vocoded/FVRA__animate__77f__vocoded__id02359_QiCI0mefCx4_00042__id03078.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps from moment to moment more than it would in a real recording. The brightness also jumps noticeably from one frame to the next. The audio sounds like a natural recording, with nothing altered.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "animate", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording", "score": "melt", "value": 1.571}, {"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.409}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face wobbles and warps from moment to moment more than it would in a real recording. The brightness also jumps noticeably from one frame to the next.", "audio": "The audio sounds like a natural recording, with nothing altered."}, "evidence": [{"modality": "video", "artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording"}, {"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "animate", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id02359", "target_id": "id03078", "pair_id": "id02359_QiCI0mefCx4_00042__id03078", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__animate__77f__vocoded__id02359_QiCI0mefCx4_00042__id03078.mp4"}, {"id": "t4_FVRA__facefusion__77f__vocoded__id08352_ANh03YiJIvE_00031__id08394", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/facefusion/77f/vocoded/FVRA__facefusion__77f__vocoded__id08352_ANh03YiJIvE_00031__id08394.mov"], "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 frame", "voice"], "gt_answer": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. Nothing in the picture looks visibly wrong. The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["inner face"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": []}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The inner face, meaning the eyes, nose, mouth and skin, is AI-generated, while the hair, head shape, body and background are untouched. Nothing in the picture looks visibly wrong.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed."}, "evidence": []}, "meta": {"video_method": "facefusion", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id08352", "target_id": "id08394", "pair_id": "id08352_ANh03YiJIvE_00031__id08394", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__facefusion__77f__vocoded__id08352_ANh03YiJIvE_00031__id08394.mp4"}, {"id": "t4_RVFA__seed_vc__77f__main__id07630_hOMvueEQFQU_00049__id08406", "task": "task4_explanation", "split": "test", "clips": ["videos/RVFA/seed_vc/77f/main/RVFA__seed_vc__77f__main__id07630_hOMvueEQFQU_00049__id08406.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural recording, with nothing altered in the face or anywhere else in the frame. The voice is AI-generated, though the video itself is untouched. Nothing in the speech sounds wrong or out of place.", "gt_label": "RVFA", "gt_target": {"fake_parts": ["audio"], "regions": ["voice"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The video looks like a natural recording, with nothing altered in the face or anywhere else in the frame.", "audio": "The voice is AI-generated, though the video itself is untouched. Nothing in the speech sounds wrong or out of place."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id07630", "target_id": "id08406", "pair_id": "id07630_hOMvueEQFQU_00049__id08406", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVFA__seed_vc__77f__main__id07630_hOMvueEQFQU_00049__id08406.mp4"}, {"id": "t4_FVFA__s2v__77f__main__id03123_UWxV82LU0Iw_00101__id04597", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/s2v/77f/main/FVFA__s2v__77f__main__id03123_UWxV82LU0Iw_00101__id04597.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The most visible sign is that the brightness jumps from one frame to the next, so the picture flickers lighter and darker instead of holding a steady exposure. The voice is AI-altered. No specific sign in the sound is pointed out.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["whole frame", "voice"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames", "score": "flicker", "value": 0.662}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The most visible sign is that the brightness jumps from one frame to the next, so the picture flickers lighter and darker instead of holding a steady exposure.", "audio": "The voice is AI-altered. No specific sign in the sound is pointed out."}, "evidence": [{"modality": "video", "artifact": "color or lighting inconsistency", "meaning": "the brightness jumps between frames"}]}, "meta": {"video_method": "s2v", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id03123", "target_id": "id04597", "pair_id": "id03123_UWxV82LU0Iw_00101__id04597", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__s2v__77f__main__id03123_UWxV82LU0Iw_00101__id04597.mp4"}, {"id": "t4_RVRA__real__77f__vocoded__id07697_diNcpyVMCk8_00047__id07630", "task": "task4_explanation", "split": "test", "clips": ["videos/RVRA/real/77f/vocoded/RVRA__real__77f__vocoded__id07697_diNcpyVMCk8_00047__id07630.mov"], "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 frame", "voice"], "gt_answer": "The video looks like a natural, unaltered recording. Movement, lighting and detail all behave the way they would in real footage, and nothing in it appears generated or changed. The audio sounds like a natural, unaltered recording. The voice and any background sound come across as real and consistent, with nothing that seems generated or altered.", "gt_label": "RVRA", "gt_target": {"fake_parts": [], "regions": ["none"], "parts": {"video": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "a camera recording"}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The video looks like a natural, unaltered recording. Movement, lighting and detail all behave the way they would in real footage, and nothing in it appears generated or changed.", "audio": "The audio sounds like a natural, unaltered recording. The voice and any background sound come across as real and consistent, with nothing that seems generated or altered."}, "evidence": []}, "meta": {"video_method": "real", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id07697", "target_id": "id07630", "pair_id": "id07697_diNcpyVMCk8_00047__id07630", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/RVRA__real__77f__vocoded__id07697_diNcpyVMCk8_00047__id07630.mp4"}, {"id": "t4_FVFA__facefusion__77f__main__id07054_zTKKYEviZvQ_00075__id01052", "task": "task4_explanation", "split": "test", "clips": ["videos/FVFA/facefusion/77f/main/FVFA__facefusion__77f__main__id07054_zTKKYEviZvQ_00075__id01052.mov"], "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 frame", "voice"], "gt_answer": "The inner face (eyes, nose, mouth and skin) is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with softer detail across the features. The voice has been altered by AI. There is no specific sign to point to beyond that.", "gt_label": "FVFA", "gt_target": {"fake_parts": ["video", "audio"], "regions": ["inner face", "voice"], "parts": {"video": {"fake": true, "region": "inner face", "method": "facefusion", "changed": "the inner face (eyes, nose, mouth, skin) is AI-generated", "original": "the hair, head shape, body and background are untouched", "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.812}]}, "audio": {"fake": true, "region": "voice", "method": "seed_vc", "changed": "the voice is AI-altered", "original": null}}, "sections": {"video": "The inner face (eyes, nose, mouth and skin) is AI-generated, while the hair, head shape, body and background are untouched. The face looks blurrier than it would in a real recording, with softer detail across the features.", "audio": "The voice has been altered by AI. There is no specific sign to point to beyond that."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}]}, "meta": {"video_method": "facefusion", "audio_method": "seed_vc", "frames": 77, "arm": "main", "source_id": "id07054", "target_id": "id01052", "pair_id": "id07054_zTKKYEviZvQ_00075__id01052", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVFA__facefusion__77f__main__id07054_zTKKYEviZvQ_00075__id01052.mp4"}, {"id": "t4_FVRA__s2v__77f__vocoded__id01858_HBSDURviflg_00004__id01156", "task": "task4_explanation", "split": "test", "clips": ["videos/FVRA/s2v/77f/vocoded/FVRA__s2v__77f__vocoded__id01858_HBSDURviflg_00004__id01156.mov"], "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 frame", "voice"], "gt_answer": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft detail throughout. It also wobbles and warps as the person moves, shifting in a way a real face on camera would not. The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed.", "gt_label": "FVRA", "gt_target": {"fake_parts": ["video"], "regions": ["whole frame"], "parts": {"video": {"fake": true, "region": "whole frame", "method": "s2v", "changed": "the whole frame, person and background, is AI-generated", "original": null, "signs": [{"artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording", "score": "sharp", "value": 0.704}, {"artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording", "score": "melt", "value": 1.474}]}, "audio": {"fake": false, "region": "none", "method": "real", "changed": null, "original": "the speaker's real voice, passed through a vocoder only"}}, "sections": {"video": "The whole frame is AI-generated, both the person and the background. The face looks blurrier than it would in a real recording, with soft detail throughout. It also wobbles and warps as the person moves, shifting in a way a real face on camera would not.", "audio": "The audio sounds like a natural, unaltered recording, with nothing in the voice or sound that has been changed."}, "evidence": [{"modality": "video", "artifact": "unnatural blur", "meaning": "the face is blurrier than in a real recording"}, {"modality": "video", "artifact": "shape or structure distortion", "meaning": "the face wobbles or warps more than in a real recording"}]}, "meta": {"video_method": "s2v", "audio_method": "bigvgan", "frames": 77, "arm": "vocoded", "source_id": "id01858", "target_id": "id01156", "pair_id": "id01858_HBSDURviflg_00004__id01156", "annotator": "claude-sonnet-5-5", "human_checked": false}, "mp4": "videos/FVRA__s2v__77f__vocoded__id01858_HBSDURviflg_00004__id01156.mp4"}]