{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:I24L2MWN7N7XIQC4GLU75DY4IV","short_pith_number":"pith:I24L2MWN","schema_version":"1.0","canonical_sha256":"46b8bd32cdfb7f74405c32e9fe8f1c457fbb721b4f7ab3f0a2022ac2b7d3464e","source":{"kind":"arxiv","id":"2608.08569","version":1},"attestation_state":"computed","paper":{"title":"VoxZip: Semantic-Anchored Temporal KV Cache Compression for Long-Context Audio Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"cs.AI","authors_text":"Dongjie Fu, Fangming Feng, Linjun Li, Tao Jin, Wenshi Chen, Wenxu Jia, Xize Cheng, Yingming Li, Zhou Zhao","submitted_at":"2026-08-09T08:17:18Z","abstract_excerpt":"Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks. Despite this progress, their long-context inference remains severely bottlenecked by prohibitive KV cache memory demands. Existing text-centric compression methods struggle here, often disrupting speech continuity or discarding crucial semantic cues. To address this, we propose VoxZip, a train-free, two-stage semantic-anchored KV cache compression framework. The first stage uses automatic speech recognition (ASR) transcriptions as explicit semantic anchors to temp"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2608.08569","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-08-09T08:17:18Z","cross_cats_sorted":["cs.SD"],"title_canon_sha256":"b01768795e4b59b26876c197eb0f117ecbf2dff3ddff26590191dae921d2dc8e","abstract_canon_sha256":"05a9a29f7100ecf2bd09d8258c4236c0fbd7d6089d40e0f401402806b72f4ff7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:23:00.551809Z","signature_b64":"9Br0wuWQiYs9Cggs/H7/A+ZD4eLSKdUFZmRFUs2FOfv9BrI1cXdH/XpeMx2qtGNpOIyIdWdxGvT82mjL7iIiCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"46b8bd32cdfb7f74405c32e9fe8f1c457fbb721b4f7ab3f0a2022ac2b7d3464e","last_reissued_at":"2026-08-11T01:23:00.549399Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:23:00.549399Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VoxZip: Semantic-Anchored Temporal KV Cache Compression for Long-Context Audio Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"cs.AI","authors_text":"Dongjie Fu, Fangming Feng, Linjun Li, Tao Jin, Wenshi Chen, Wenxu Jia, Xize Cheng, Yingming Li, Zhou Zhao","submitted_at":"2026-08-09T08:17:18Z","abstract_excerpt":"Recent advancements in Speech Large Language Models have demonstrated remarkable capabilities in understanding complex audio tasks. Despite this progress, their long-context inference remains severely bottlenecked by prohibitive KV cache memory demands. Existing text-centric compression methods struggle here, often disrupting speech continuity or discarding crucial semantic cues. To address this, we propose VoxZip, a train-free, two-stage semantic-anchored KV cache compression framework. The first stage uses automatic speech recognition (ASR) transcriptions as explicit semantic anchors to temp"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.08569","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2608.08569/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"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":"2608.08569","created_at":"2026-08-11T01:23:00.550148+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.08569v1","created_at":"2026-08-11T01:23:00.550148+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.08569","created_at":"2026-08-11T01:23:00.550148+00:00"},{"alias_kind":"pith_short_12","alias_value":"I24L2MWN7N7X","created_at":"2026-08-11T01:23:00.550148+00:00"},{"alias_kind":"pith_short_16","alias_value":"I24L2MWN7N7XIQC4","created_at":"2026-08-11T01:23:00.550148+00:00"},{"alias_kind":"pith_short_8","alias_value":"I24L2MWN","created_at":"2026-08-11T01:23:00.550148+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/I24L2MWN7N7XIQC4GLU75DY4IV","json":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV.json","graph_json":"https://pith.science/api/pith-number/I24L2MWN7N7XIQC4GLU75DY4IV/graph.json","events_json":"https://pith.science/api/pith-number/I24L2MWN7N7XIQC4GLU75DY4IV/events.json","paper":"https://pith.science/paper/I24L2MWN"},"agent_actions":{"view_html":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV","download_json":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV.json","view_paper":"https://pith.science/paper/I24L2MWN","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.08569&json=true","fetch_graph":"https://pith.science/api/pith-number/I24L2MWN7N7XIQC4GLU75DY4IV/graph.json","fetch_events":"https://pith.science/api/pith-number/I24L2MWN7N7XIQC4GLU75DY4IV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV/action/storage_attestation","attest_author":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV/action/author_attestation","sign_citation":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV/action/citation_signature","submit_replication":"https://pith.science/pith/I24L2MWN7N7XIQC4GLU75DY4IV/action/replication_record"}},"created_at":"2026-08-11T01:23:00.550148+00:00","updated_at":"2026-08-11T01:23:00.550148+00:00"}