{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:7N46RRKUTD7FJHAT2W3ASSRGG3","short_pith_number":"pith:7N46RRKU","schema_version":"1.0","canonical_sha256":"fb79e8c55498fe549c13d5b6094a2636f2460109742fc275f69b3122d9c23a49","source":{"kind":"arxiv","id":"2505.04457","version":4},"attestation_state":"computed","paper":{"title":"Miipher-2: A Universal Speech Restoration Model for Million-Hour Scale Data Restoration","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","eess.AS"],"primary_cat":"cs.SD","authors_text":"Haruko Ishikawa, Heiga Zen, Michiel Bacchiani, Robin Scheibler, Shigeki Karita, Yuma Koizumi","submitted_at":"2025-05-07T14:27:46Z","abstract_excerpt":"Training data cleaning is a new application for generative model-based speech restoration (SR). This paper introduces Miipher-2, an SR model designed for million-hour scale data, for training data cleaning for large-scale generative models like large language models. Key challenges addressed include generalization to unseen languages, operation without explicit conditioning (e.g., text, speaker ID), and computational efficiency. Miipher-2 utilizes a frozen, pre-trained Universal Speech Model (USM), supporting over 300 languages, as a robust, conditioning-free feature extractor. To optimize eff"},"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":"2505.04457","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SD","submitted_at":"2025-05-07T14:27:46Z","cross_cats_sorted":["cs.CL","eess.AS"],"title_canon_sha256":"772a6f935d83054d61b1c5b92021af58b4e6e3928a542d6ab943cd5739879bdb","abstract_canon_sha256":"6ee8b968aa4594ad7eab7da7a5548732a0efcbebd1d2b806d65ed6241e5eb9ac"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:41:41.698419Z","signature_b64":"vnuwulh2mmYD/uPdRt2oKoMWSGH4/4RsSbnq8aXELFAHRK9312ptsa/xNSZ7VRFkjUdWJ318yshVW8ILegWgCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fb79e8c55498fe549c13d5b6094a2636f2460109742fc275f69b3122d9c23a49","last_reissued_at":"2026-07-05T11:41:41.697877Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:41:41.697877Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Miipher-2: A Universal Speech Restoration Model for Million-Hour Scale Data Restoration","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","eess.AS"],"primary_cat":"cs.SD","authors_text":"Haruko Ishikawa, Heiga Zen, Michiel Bacchiani, Robin Scheibler, Shigeki Karita, Yuma Koizumi","submitted_at":"2025-05-07T14:27:46Z","abstract_excerpt":"Training data cleaning is a new application for generative model-based speech restoration (SR). This paper introduces Miipher-2, an SR model designed for million-hour scale data, for training data cleaning for large-scale generative models like large language models. Key challenges addressed include generalization to unseen languages, operation without explicit conditioning (e.g., text, speaker ID), and computational efficiency. Miipher-2 utilizes a frozen, pre-trained Universal Speech Model (USM), supporting over 300 languages, as a robust, conditioning-free feature extractor. To optimize eff"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.04457","kind":"arxiv","version":4},"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/2505.04457/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":"2505.04457","created_at":"2026-07-05T11:41:41.697952+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.04457v4","created_at":"2026-07-05T11:41:41.697952+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.04457","created_at":"2026-07-05T11:41:41.697952+00:00"},{"alias_kind":"pith_short_12","alias_value":"7N46RRKUTD7F","created_at":"2026-07-05T11:41:41.697952+00:00"},{"alias_kind":"pith_short_16","alias_value":"7N46RRKUTD7FJHAT","created_at":"2026-07-05T11:41:41.697952+00:00"},{"alias_kind":"pith_short_8","alias_value":"7N46RRKU","created_at":"2026-07-05T11:41:41.697952+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2603.02641","citing_title":"Rethinking Training Targets, Architectures and Data Quality for Universal Speech Enhancement","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3","json":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3.json","graph_json":"https://pith.science/api/pith-number/7N46RRKUTD7FJHAT2W3ASSRGG3/graph.json","events_json":"https://pith.science/api/pith-number/7N46RRKUTD7FJHAT2W3ASSRGG3/events.json","paper":"https://pith.science/paper/7N46RRKU"},"agent_actions":{"view_html":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3","download_json":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3.json","view_paper":"https://pith.science/paper/7N46RRKU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.04457&json=true","fetch_graph":"https://pith.science/api/pith-number/7N46RRKUTD7FJHAT2W3ASSRGG3/graph.json","fetch_events":"https://pith.science/api/pith-number/7N46RRKUTD7FJHAT2W3ASSRGG3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3/action/storage_attestation","attest_author":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3/action/author_attestation","sign_citation":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3/action/citation_signature","submit_replication":"https://pith.science/pith/7N46RRKUTD7FJHAT2W3ASSRGG3/action/replication_record"}},"created_at":"2026-07-05T11:41:41.697952+00:00","updated_at":"2026-07-05T11:41:41.697952+00:00"}