{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.26244", "kind": "paper", "url": "https://arxiv.org/abs/2605.26244"}, {"id": "gh:pkucs-ltf/longav-compass", "kind": "repo", "url": "https://github.com/pkucs-Ltf/LongAV-Compass"}], "authors": ["Tengfei Liu", "Yang Shi", "Xuanyu Zhu", "Jiafu Tang", "Liu Yang", "Qixun Wang", "Zhuoran Zhang", "Yuqi Tang", "Fengxiang Wang", "Yuhao Dong", "Xinlong Chen", "Bozhou Li", "Bohan Zeng", "Yue Ding", "Xiaohan Zhang", "Jialu Chen", "Haotian Wang", "Yuanxing Zhang", "Pengfei Wan", "Leye Wang"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.26244", "source_url": "https://arxiv.org/abs/2605.26244"}, "categories": ["General AI", "Multimodal", "cs.CV"], "description": {"en": "LongAV-Compass evaluates minute-scale audio-visual generation across text-to-audio-video, image-to-audio-video, and video-to-audio-video tasks. It includes 284 curated test cases and an evaluation framework combining multimodal metrics with MLLM-assisted assessment across 20+ dimensions covering segment quality, cross-segment consistency, narrative coherence, semantic alignment, and audiovisual synchronization."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.26244", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.26244", "title": "LongAV-Compass"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Peking University", "Nanjing University", "Shanghai Jiao Tong University", "Hong Kong University of Science and Technology", "The Hong Kong University of Science and Technology (Guangzhou)", "Shanghai Artificial Intelligence Laboratory", "Nanyang Technological University", "Institute of Automation", "Tsinghua University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2605.26244", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2605.26244", "source_url": "https://openalex.org/W7162556597"}, "key": "claire-radar:2605.26244", "modality": null, "name": "LongAV-Compass", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_longav-compass_f7e880eb", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "8e335dac04cc34850b30dbcef2555721d546835c0e8bd2b3b480b10c12919473", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.26244"}, "publisher": null, "released": "2026-05-25", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.26244", "source_url": "https://arxiv.org/abs/2605.26244"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-26244", "source": "claire_radar", "source_benchmark_id": "2605.26244", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Multimodal", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/pkucs-Ltf/LongAV-Compass", "githubScope": "benchmark_repo", "githubStars": 17, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": 36, "hfPaperUrl": "https://huggingface.co/papers/2605.26244", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "lastSuccessfulDate": "2026-09-30", "state": "stale"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "903d8a598d71a3ad64fe4c5d7fe6a2514e7320ca4ad22b1e9d5c1061ad0eb1d6", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "LongAV-Compass evaluates minute-scale audio-visual generation across text-to-audio-video, image-to-audio-video, and video-to-audio-video tasks. It includes 284 curated test cases and an evaluation framework combining multimodal metrics with MLLM-assisted assessment across 20+ dimensions covering segment quality, cross-segment consistency, narrative coherence, semantic alignment, and audiovisual synchronization.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["exact coined title identity tied to benchmark evidence", "benchmark release stated in one sentence"], "snippet": "To bridge this gap, we introduce LongAV-Compass, a systematic benchmark for minute-long audio-visual generation."}, "familyId": "bmf_67cc07e08a0c", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/pkucs-Ltf/LongAV-Compass", "url": "https://github.com/pkucs-Ltf/LongAV-Compass"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/pkucs-Ltf/LongAV-Compass", "url": "https://huggingface.co/papers/2605.26244"}], "scope": "benchmark_repo", "url": "https://github.com/pkucs-Ltf/LongAV-Compass", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_longav-compass_f7e880eb", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/pkucs-Ltf/LongAV-Compass", "data": null, "hfPaper": "https://huggingface.co/papers/2605.26244", "paper": "https://arxiv.org/abs/2605.26244", "pdf": "https://arxiv.org/pdf/2605.26244", "project": null, "report": "https://arxiv.org/abs/2605.26244"}, "metrics": [], "motivation": "Audio-visual generation is rapidly advancing from short clips to minute-long content, while existing evaluation protocols remain largely confined to short-form settings.", "name": "LongAV-Compass", "oneLine": "LongAV-Compass evaluates minute-scale audio-visual generation across text-to-audio-video, image-to-audio-video, and video-to-audio-video tasks. It includes 284 curated test cases and an evaluation framework combining multimodal metrics with MLLM-assisted assessment across 20+ dimensions covering segment quality, cross-segment consistency, narrative coherence, semantic alignment, and audiovisual synchronization.", "paperTitle": "LongAV-Compass: Towards Unified Evaluation of Minute-Scale Audio-Visual Generation Across T2AV, I2AV, and V2AV", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.26244", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "PKU CS LongAV Group", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/pkucs-Ltf/LongAV-Compass"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-05-25", "source": {"authors": ["Tengfei Liu", "Yang Shi", "Xuanyu Zhu", "Jiafu Tang", "Liu Yang", "Qixun Wang", "Zhuoran Zhang", "Yuqi Tang", "Fengxiang Wang", "Yuhao Dong", "Xinlong Chen", "Bozhou Li", "Bohan Zeng", "Yue Ding", "Xiaohan Zhang", "Jialu Chen", "Haotian Wang", "Yuanxing Zhang", "Pengfei Wan", "Leye Wang"], "categories": ["cs.CV", "cs.MM", "cs.SD"], "id": "2605.26244", "title": "LongAV-Compass: Towards Unified Evaluation of Minute-Scale Audio-Visual Generation Across T2AV, I2AV, and V2AV", "type": "arxiv", "url": "https://arxiv.org/abs/2605.26244"}, "sourceUpdatedAt": "2026-05-25T18:12:09Z", "topics": ["cs.CV"], "venueAttempts": [], "whyItMatters": "Existing audio-visual evaluation is limited to short clips, leaving a gap for minute-scale generation. LongAV-Compass provides a standardized protocol for diagnosing degradation in identity consistency, narrative coherence, and alignment over long horizons, supporting comparison of long-form generation models."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
