{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:YBAQQSACI3W57PZFOKT4KJT55G","short_pith_number":"pith:YBAQQSAC","schema_version":"1.0","canonical_sha256":"c04108480246eddfbf2572a7c5267de9a2b11f99a161643cecf790ec76526185","source":{"kind":"arxiv","id":"2509.09197","version":1},"attestation_state":"computed","paper":{"title":"Improving Synthetic Data Training for Contextual Biasing Models with a Keyword-Aware Cost Function","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chin Yuen Kwok, Eng Siong Chng, Jia Qi Yip","submitted_at":"2025-09-11T07:12:17Z","abstract_excerpt":"Rare word recognition can be improved by adapting ASR models to synthetic data that includes these words. Further improvements can be achieved through contextual biasing, which trains and adds a biasing module into the model architecture to prioritize rare words. While training the module on synthetic rare word data is more effective than using non-rare-word data, it can lead to overfitting due to artifacts in the synthetic audio. To address this, we enhance the TCPGen-based contextual biasing approach and propose a keyword-aware loss function that additionally focuses on biased words when tra"},"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":"2509.09197","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-09-11T07:12:17Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"fda4a61f766139692a9b07d0778ddc447c0f61f15edf58fdd40e58802ec98143","abstract_canon_sha256":"c345a270c92bbe8e57e8882c638190d6ddd6029ca94152384a4043f627e7b29a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:09:31.455995Z","signature_b64":"whjc83nWMVG7B7JZXS14jUsuuRI/RhVI3vl6FzNAO79bYP0c3jrhpNIF7uyvI0P7xhtoIO8jcWBoewqFbW67Cg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c04108480246eddfbf2572a7c5267de9a2b11f99a161643cecf790ec76526185","last_reissued_at":"2026-07-05T12:09:31.455473Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:09:31.455473Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Improving Synthetic Data Training for Contextual Biasing Models with a Keyword-Aware Cost Function","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Chin Yuen Kwok, Eng Siong Chng, Jia Qi Yip","submitted_at":"2025-09-11T07:12:17Z","abstract_excerpt":"Rare word recognition can be improved by adapting ASR models to synthetic data that includes these words. Further improvements can be achieved through contextual biasing, which trains and adds a biasing module into the model architecture to prioritize rare words. While training the module on synthetic rare word data is more effective than using non-rare-word data, it can lead to overfitting due to artifacts in the synthetic audio. To address this, we enhance the TCPGen-based contextual biasing approach and propose a keyword-aware loss function that additionally focuses on biased words when tra"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.09197","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/2509.09197/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":"2509.09197","created_at":"2026-07-05T12:09:31.455531+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.09197v1","created_at":"2026-07-05T12:09:31.455531+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.09197","created_at":"2026-07-05T12:09:31.455531+00:00"},{"alias_kind":"pith_short_12","alias_value":"YBAQQSACI3W5","created_at":"2026-07-05T12:09:31.455531+00:00"},{"alias_kind":"pith_short_16","alias_value":"YBAQQSACI3W57PZF","created_at":"2026-07-05T12:09:31.455531+00:00"},{"alias_kind":"pith_short_8","alias_value":"YBAQQSAC","created_at":"2026-07-05T12:09:31.455531+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/YBAQQSACI3W57PZFOKT4KJT55G","json":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G.json","graph_json":"https://pith.science/api/pith-number/YBAQQSACI3W57PZFOKT4KJT55G/graph.json","events_json":"https://pith.science/api/pith-number/YBAQQSACI3W57PZFOKT4KJT55G/events.json","paper":"https://pith.science/paper/YBAQQSAC"},"agent_actions":{"view_html":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G","download_json":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G.json","view_paper":"https://pith.science/paper/YBAQQSAC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.09197&json=true","fetch_graph":"https://pith.science/api/pith-number/YBAQQSACI3W57PZFOKT4KJT55G/graph.json","fetch_events":"https://pith.science/api/pith-number/YBAQQSACI3W57PZFOKT4KJT55G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G/action/storage_attestation","attest_author":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G/action/author_attestation","sign_citation":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G/action/citation_signature","submit_replication":"https://pith.science/pith/YBAQQSACI3W57PZFOKT4KJT55G/action/replication_record"}},"created_at":"2026-07-05T12:09:31.455531+00:00","updated_at":"2026-07-05T12:09:31.455531+00:00"}