{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:KRX662XDVC7Y7TE7Q4IDKQJK6V","short_pith_number":"pith:KRX662XD","schema_version":"1.0","canonical_sha256":"546fef6ae3a8bf8fcc9f871035412af55d522aacddd417ff6bcd8ac7fcc55f85","source":{"kind":"arxiv","id":"2606.13507","version":1},"attestation_state":"computed","paper":{"title":"Leveraging Audio-LLMs to Filter Speech-to-Speech Training Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Qixu Chen, Satoshi Nakamura","submitted_at":"2026-06-11T15:55:23Z","abstract_excerpt":"Large-scale mined corpora provide abundant training data for end-to-end speech-to-speech translation (S2ST) but may contain noise, misalignment, and semantic errors. Filtering noisy data is crucial to maintain robust speech translation performance. We study how to train an audio-language model to make keep/drop decisions on paired speech directly from audio. To obtain reliable supervision without manual labels, we adopt a scalable two-stage Rank-to-Distill strategy. A lightweight ranker generates keep/drop pseudo-labels from noisy speech pairs, then trains an audio large language model to pred"},"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":"2606.13507","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-06-11T15:55:23Z","cross_cats_sorted":[],"title_canon_sha256":"b462dd2589e4aeaa70487db88e21445f92e4f8521777c2c2852771e9de6cc659","abstract_canon_sha256":"7b973eec6a966459691f837557068620f16b07d9a8f9822a9fa96269edf2253b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-12T01:10:07.382762Z","signature_b64":"wZOPQmiA7RKlysVJyipCO9uItuXkVYobrY+J69wcFyxDX7+1aKut2CIXLGoCEvH4T/9mSN9LxIGUOkkcuOaoAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"546fef6ae3a8bf8fcc9f871035412af55d522aacddd417ff6bcd8ac7fcc55f85","last_reissued_at":"2026-06-12T01:10:07.381866Z","signature_status":"signed_v1","first_computed_at":"2026-06-12T01:10:07.381866Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Leveraging Audio-LLMs to Filter Speech-to-Speech Training Data","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Qixu Chen, Satoshi Nakamura","submitted_at":"2026-06-11T15:55:23Z","abstract_excerpt":"Large-scale mined corpora provide abundant training data for end-to-end speech-to-speech translation (S2ST) but may contain noise, misalignment, and semantic errors. Filtering noisy data is crucial to maintain robust speech translation performance. We study how to train an audio-language model to make keep/drop decisions on paired speech directly from audio. To obtain reliable supervision without manual labels, we adopt a scalable two-stage Rank-to-Distill strategy. A lightweight ranker generates keep/drop pseudo-labels from noisy speech pairs, then trains an audio large language model to pred"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.13507","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/2606.13507/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":"2606.13507","created_at":"2026-06-12T01:10:07.382027+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.13507v1","created_at":"2026-06-12T01:10:07.382027+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.13507","created_at":"2026-06-12T01:10:07.382027+00:00"},{"alias_kind":"pith_short_12","alias_value":"KRX662XDVC7Y","created_at":"2026-06-12T01:10:07.382027+00:00"},{"alias_kind":"pith_short_16","alias_value":"KRX662XDVC7Y7TE7","created_at":"2026-06-12T01:10:07.382027+00:00"},{"alias_kind":"pith_short_8","alias_value":"KRX662XD","created_at":"2026-06-12T01:10:07.382027+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2606.13507","citing_title":"Leveraging Audio-LLMs to Filter Speech-to-Speech Training Data","ref_index":3,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V","json":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V.json","graph_json":"https://pith.science/api/pith-number/KRX662XDVC7Y7TE7Q4IDKQJK6V/graph.json","events_json":"https://pith.science/api/pith-number/KRX662XDVC7Y7TE7Q4IDKQJK6V/events.json","paper":"https://pith.science/paper/KRX662XD"},"agent_actions":{"view_html":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V","download_json":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V.json","view_paper":"https://pith.science/paper/KRX662XD","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.13507&json=true","fetch_graph":"https://pith.science/api/pith-number/KRX662XDVC7Y7TE7Q4IDKQJK6V/graph.json","fetch_events":"https://pith.science/api/pith-number/KRX662XDVC7Y7TE7Q4IDKQJK6V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V/action/storage_attestation","attest_author":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V/action/author_attestation","sign_citation":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V/action/citation_signature","submit_replication":"https://pith.science/pith/KRX662XDVC7Y7TE7Q4IDKQJK6V/action/replication_record"}},"created_at":"2026-06-12T01:10:07.382027+00:00","updated_at":"2026-06-12T01:10:07.382027+00:00"}