{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QSKAJ67DCH2K5MRRRTKUWQNYGF","short_pith_number":"pith:QSKAJ67D","schema_version":"1.0","canonical_sha256":"849404fbe311f4aeb2318cd54b41b83177f1058cb606d3c3f6065a2e5380f23b","source":{"kind":"arxiv","id":"2512.03637","version":2},"attestation_state":"computed","paper":{"title":"AaSP: Aliasing-aware Self-Supervised Pre-Training for Audio Spectrogram Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"AaSP uses input-estimated kernels to fuse high-frequency cues lost to aliasing in audio spectrogram pre-training.","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.SD","authors_text":"Kohei Yamamoto, Kosuke Okusa","submitted_at":"2025-12-03T10:17:35Z","abstract_excerpt":"Transformer-based audio self-supervised learning (SSL) models commonly use spectrograms, vision-style Transformers, and masked modeling objectives. However, convolutional patchification with temporal downsampling lowers the effective Nyquist frequency and introduces aliasing, while na\\\"ive low-pass filtering may remove task-relevant high-frequency cues. We present AaSP, an aliasing-aware self-supervised pre-training framework for audio spectrogram transformers. AaSP combines an aliasing-aware patch representation, teacher-student masked modeling, a cross-attention predictor, and multi-mask con"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2512.03637","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SD","submitted_at":"2025-12-03T10:17:35Z","cross_cats_sorted":["cs.LG","stat.ML"],"title_canon_sha256":"459598cee3f40ec87ef99209c8389e007e070721f92715f7544fc1d810b56584","abstract_canon_sha256":"35cb764cb3457dc05a500bd8f4fa4042cc9f06e2cc683b35baa12ca56633f960"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:16.929718Z","signature_b64":"MbhtdaOrwuzmqb6F9RsKBZJTSP4zIW7IOJQvZVG55HW7wQ0RR/Ve8icDWK5Jf7uRMBWEPGqpyMHFHFoo0JKUBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"849404fbe311f4aeb2318cd54b41b83177f1058cb606d3c3f6065a2e5380f23b","last_reissued_at":"2026-05-17T23:39:16.928935Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:16.928935Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"AaSP: Aliasing-aware Self-Supervised Pre-Training for Audio Spectrogram Transformers","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"AaSP uses input-estimated kernels to fuse high-frequency cues lost to aliasing in audio spectrogram pre-training.","cross_cats":["cs.LG","stat.ML"],"primary_cat":"cs.SD","authors_text":"Kohei Yamamoto, Kosuke Okusa","submitted_at":"2025-12-03T10:17:35Z","abstract_excerpt":"Transformer-based audio self-supervised learning (SSL) models commonly use spectrograms, vision-style Transformers, and masked modeling objectives. However, convolutional patchification with temporal downsampling lowers the effective Nyquist frequency and introduces aliasing, while na\\\"ive low-pass filtering may remove task-relevant high-frequency cues. We present AaSP, an aliasing-aware self-supervised pre-training framework for audio spectrogram transformers. AaSP combines an aliasing-aware patch representation, teacher-student masked modeling, a cross-attention predictor, and multi-mask con"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Under fine-tuning, the full AaSP framework achieves state-of-the-art results on AS-20K, ESC-50, and NSynth among compared self-supervised baselines, while remaining competitive elsewhere.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the aliasing introduced by convolutional patchification with temporal downsampling is a primary performance bottleneck and that fusing features from alias-prone modulation bands via an input-estimated kernel will integrate useful high-frequency cues without introducing new instabilities or overfitting to the estimation process.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"AaSP learns aliasing-stable audio representations by augmenting patch tokens with adaptive subband features from alias-prone bands and using teacher-student masked modeling plus multi-mask contrastive regularization, reaching SOTA on AS-20K, ESC-50, and NSynth under fine-tuning.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"AaSP uses input-estimated kernels to fuse high-frequency cues lost to aliasing in audio spectrogram pre-training.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"e2fb0eeed33726c91a3e1e815de7f7b3c468d9266dc1ce4d879765791b69d827"},"source":{"id":"2512.03637","kind":"arxiv","version":2},"verdict":{"id":"80797920-5490-401d-9a58-108f8acb1b2a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T02:26:53.341874Z","strongest_claim":"Under fine-tuning, the full AaSP framework achieves state-of-the-art results on AS-20K, ESC-50, and NSynth among compared self-supervised baselines, while remaining competitive elsewhere.","one_line_summary":"AaSP learns aliasing-stable audio representations by augmenting patch tokens with adaptive subband features from alias-prone bands and using teacher-student masked modeling plus multi-mask contrastive regularization, reaching SOTA on AS-20K, ESC-50, and NSynth under fine-tuning.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the aliasing introduced by convolutional patchification with temporal downsampling is a primary performance bottleneck and that fusing features from alias-prone modulation bands via an input-estimated kernel will integrate useful high-frequency cues without introducing new instabilities or overfitting to the estimation process.","pith_extraction_headline":"AaSP uses input-estimated kernels to fuse high-frequency cues lost to aliasing in audio spectrogram pre-training."},"references":{"count":45,"sample":[{"doi":"","year":2019,"title":"Bert: Pre- training of deep bidirectional transformers for language understanding,","work_id":"3344a935-5043-4695-b583-1a9d4f835740","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Dinov2: Learning robust visual features without supervision,","work_id":"f7c73b19-2235-4783-b6e2-f3a6aef0e8ee","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Masked autoencoders that listen,","work_id":"ee7a0027-5933-4c94-909a-47476d7a5b0d","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"MAE-AST: Masked Autoencoding Audio Spectrogram Transformer,","work_id":"d8738218-0d97-4b38-bc5f-f0863e4dea54","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"SSAST: Self-Supervised Audio Spectrogram Transformer,","work_id":"70a7693f-62ad-4e08-bb28-3d8b3455d29b","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":45,"snapshot_sha256":"c5f8e7b3da89fc6f9329169dba6932c50d28b952972d22e663b3b6868b5cfafd","internal_anchors":3},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2512.03637","created_at":"2026-05-17T23:39:16.929056+00:00"},{"alias_kind":"arxiv_version","alias_value":"2512.03637v2","created_at":"2026-05-17T23:39:16.929056+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.03637","created_at":"2026-05-17T23:39:16.929056+00:00"},{"alias_kind":"pith_short_12","alias_value":"QSKAJ67DCH2K","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"QSKAJ67DCH2K5MRR","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"QSKAJ67D","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF","json":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF.json","graph_json":"https://pith.science/api/pith-number/QSKAJ67DCH2K5MRRRTKUWQNYGF/graph.json","events_json":"https://pith.science/api/pith-number/QSKAJ67DCH2K5MRRRTKUWQNYGF/events.json","paper":"https://pith.science/paper/QSKAJ67D"},"agent_actions":{"view_html":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF","download_json":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF.json","view_paper":"https://pith.science/paper/QSKAJ67D","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2512.03637&json=true","fetch_graph":"https://pith.science/api/pith-number/QSKAJ67DCH2K5MRRRTKUWQNYGF/graph.json","fetch_events":"https://pith.science/api/pith-number/QSKAJ67DCH2K5MRRRTKUWQNYGF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF/action/storage_attestation","attest_author":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF/action/author_attestation","sign_citation":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF/action/citation_signature","submit_replication":"https://pith.science/pith/QSKAJ67DCH2K5MRRRTKUWQNYGF/action/replication_record"}},"created_at":"2026-05-17T23:39:16.929056+00:00","updated_at":"2026-05-17T23:39:16.929056+00:00"}