{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:5QZIMBMAD7UXJ5RUUWUUGZ65SF","short_pith_number":"pith:5QZIMBMA","schema_version":"1.0","canonical_sha256":"ec328605801fe974f634a5a94367dd9150c64b574d7d5638b1875a6d6697b326","source":{"kind":"arxiv","id":"2211.03541","version":2},"attestation_state":"computed","paper":{"title":"Multi-blank Transducers for Speech Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SD"],"primary_cat":"eess.AS","authors_text":"Boris Ginsburg, Fei Jia, Hainan Xu, Shinji Watanabe, Somshubra Majumdar","submitted_at":"2022-11-04T16:24:46Z","abstract_excerpt":"This paper proposes a modification to RNN-Transducer (RNN-T) models for automatic speech recognition (ASR). In standard RNN-T, the emission of a blank symbol consumes exactly one input frame; in our proposed method, we introduce additional blank symbols, which consume two or more input frames when emitted. We refer to the added symbols as big blanks, and the method multi-blank RNN-T. For training multi-blank RNN-Ts, we propose a novel logit under-normalization method in order to prioritize emissions of big blanks. With experiments on multiple languages and datasets, we show that multi-blank RN"},"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":"2211.03541","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.AS","submitted_at":"2022-11-04T16:24:46Z","cross_cats_sorted":["cs.LG","cs.SD"],"title_canon_sha256":"02f41e02e87bd888e958f5cd0e964596b0617b20fce957badfd14c5f150fb840","abstract_canon_sha256":"0084a019de53732d8efb3cd5133644aaa8b6e442824a8693bc85eca73ffc25e7"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:07:08.187382Z","signature_b64":"ThADtQWfET3+PRkSagXivCDIvcsJ0/ybikr0+7ErX8XI61X7Vl1ssXds6lp93P66oKsSk+PGPUyGeuzpHyfpAw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ec328605801fe974f634a5a94367dd9150c64b574d7d5638b1875a6d6697b326","last_reissued_at":"2026-07-05T08:07:08.186903Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:07:08.186903Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Multi-blank Transducers for Speech Recognition","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SD"],"primary_cat":"eess.AS","authors_text":"Boris Ginsburg, Fei Jia, Hainan Xu, Shinji Watanabe, Somshubra Majumdar","submitted_at":"2022-11-04T16:24:46Z","abstract_excerpt":"This paper proposes a modification to RNN-Transducer (RNN-T) models for automatic speech recognition (ASR). In standard RNN-T, the emission of a blank symbol consumes exactly one input frame; in our proposed method, we introduce additional blank symbols, which consume two or more input frames when emitted. We refer to the added symbols as big blanks, and the method multi-blank RNN-T. For training multi-blank RNN-Ts, we propose a novel logit under-normalization method in order to prioritize emissions of big blanks. With experiments on multiple languages and datasets, we show that multi-blank RN"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2211.03541","kind":"arxiv","version":2},"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/2211.03541/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":"2211.03541","created_at":"2026-07-05T08:07:08.186963+00:00"},{"alias_kind":"arxiv_version","alias_value":"2211.03541v2","created_at":"2026-07-05T08:07:08.186963+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.03541","created_at":"2026-07-05T08:07:08.186963+00:00"},{"alias_kind":"pith_short_12","alias_value":"5QZIMBMAD7UX","created_at":"2026-07-05T08:07:08.186963+00:00"},{"alias_kind":"pith_short_16","alias_value":"5QZIMBMAD7UXJ5RU","created_at":"2026-07-05T08:07:08.186963+00:00"},{"alias_kind":"pith_short_8","alias_value":"5QZIMBMA","created_at":"2026-07-05T08:07:08.186963+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/5QZIMBMAD7UXJ5RUUWUUGZ65SF","json":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF.json","graph_json":"https://pith.science/api/pith-number/5QZIMBMAD7UXJ5RUUWUUGZ65SF/graph.json","events_json":"https://pith.science/api/pith-number/5QZIMBMAD7UXJ5RUUWUUGZ65SF/events.json","paper":"https://pith.science/paper/5QZIMBMA"},"agent_actions":{"view_html":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF","download_json":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF.json","view_paper":"https://pith.science/paper/5QZIMBMA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2211.03541&json=true","fetch_graph":"https://pith.science/api/pith-number/5QZIMBMAD7UXJ5RUUWUUGZ65SF/graph.json","fetch_events":"https://pith.science/api/pith-number/5QZIMBMAD7UXJ5RUUWUUGZ65SF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF/action/storage_attestation","attest_author":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF/action/author_attestation","sign_citation":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF/action/citation_signature","submit_replication":"https://pith.science/pith/5QZIMBMAD7UXJ5RUUWUUGZ65SF/action/replication_record"}},"created_at":"2026-07-05T08:07:08.186963+00:00","updated_at":"2026-07-05T08:07:08.186963+00:00"}