{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:5HJF6FA5GRRGOM4OKY6GLWS6D7","short_pith_number":"pith:5HJF6FA5","schema_version":"1.0","canonical_sha256":"e9d25f141d346267338e563c65da5e1fd0339f6b67fcddc3db5cb4422b4908e0","source":{"kind":"arxiv","id":"2506.13396","version":2},"attestation_state":"computed","paper":{"title":"Bi-directional Context-Enhanced Speech Large Language Models for Multilingual Conversational ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Eng Siong Chng, Hexin Liu, Yizhou Peng","submitted_at":"2025-06-16T12:03:23Z","abstract_excerpt":"This paper introduces the integration of language-specific bi-directional context into a speech large language model (SLLM) to improve multilingual continuous conversational automatic speech recognition (ASR). We propose a character-level contextual masking strategy during training, which randomly removes portions of the context to enhance robustness and better emulate the flawed transcriptions that may occur during inference. For decoding, a two-stage pipeline is utilized: initial isolated segment decoding followed by context-aware re-decoding using neighboring hypotheses. Evaluated on the 15"},"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":"2506.13396","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-06-16T12:03:23Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"ce5e76982e25562e60d60ee93baca1a741675d913ce193fd9f7890690eeb29df","abstract_canon_sha256":"1d6f5b9c3cb4b69f6cce126f735e786dc3e777eb9a68d879e895c2c2c4c225a3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:31:46.312027Z","signature_b64":"MArCg6l4uTMcYwAmp4zK2jIIJQDROlQQs+bPN9fc5fkDTuBYeYLvejpzjWPakY+Bd/mNFWgHxdJoxgK7MbXyAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e9d25f141d346267338e563c65da5e1fd0339f6b67fcddc3db5cb4422b4908e0","last_reissued_at":"2026-07-05T11:31:46.311365Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:31:46.311365Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bi-directional Context-Enhanced Speech Large Language Models for Multilingual Conversational ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Eng Siong Chng, Hexin Liu, Yizhou Peng","submitted_at":"2025-06-16T12:03:23Z","abstract_excerpt":"This paper introduces the integration of language-specific bi-directional context into a speech large language model (SLLM) to improve multilingual continuous conversational automatic speech recognition (ASR). We propose a character-level contextual masking strategy during training, which randomly removes portions of the context to enhance robustness and better emulate the flawed transcriptions that may occur during inference. For decoding, a two-stage pipeline is utilized: initial isolated segment decoding followed by context-aware re-decoding using neighboring hypotheses. Evaluated on the 15"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.13396","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/2506.13396/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":"2506.13396","created_at":"2026-07-05T11:31:46.311445+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.13396v2","created_at":"2026-07-05T11:31:46.311445+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.13396","created_at":"2026-07-05T11:31:46.311445+00:00"},{"alias_kind":"pith_short_12","alias_value":"5HJF6FA5GRRG","created_at":"2026-07-05T11:31:46.311445+00:00"},{"alias_kind":"pith_short_16","alias_value":"5HJF6FA5GRRGOM4O","created_at":"2026-07-05T11:31:46.311445+00:00"},{"alias_kind":"pith_short_8","alias_value":"5HJF6FA5","created_at":"2026-07-05T11:31:46.311445+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.13396","citing_title":"Bi-directional Context-Enhanced Speech Large Language Models for Multilingual Conversational ASR","ref_index":2,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7","json":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7.json","graph_json":"https://pith.science/api/pith-number/5HJF6FA5GRRGOM4OKY6GLWS6D7/graph.json","events_json":"https://pith.science/api/pith-number/5HJF6FA5GRRGOM4OKY6GLWS6D7/events.json","paper":"https://pith.science/paper/5HJF6FA5"},"agent_actions":{"view_html":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7","download_json":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7.json","view_paper":"https://pith.science/paper/5HJF6FA5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.13396&json=true","fetch_graph":"https://pith.science/api/pith-number/5HJF6FA5GRRGOM4OKY6GLWS6D7/graph.json","fetch_events":"https://pith.science/api/pith-number/5HJF6FA5GRRGOM4OKY6GLWS6D7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7/action/storage_attestation","attest_author":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7/action/author_attestation","sign_citation":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7/action/citation_signature","submit_replication":"https://pith.science/pith/5HJF6FA5GRRGOM4OKY6GLWS6D7/action/replication_record"}},"created_at":"2026-07-05T11:31:46.311445+00:00","updated_at":"2026-07-05T11:31:46.311445+00:00"}