{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:N3ENU2IHEMXUSD7LFBUR2W3V2J","short_pith_number":"pith:N3ENU2IH","schema_version":"1.0","canonical_sha256":"6ec8da6907232f490feb28691d5b75d254bec7d1a52b94f3e9ebaefa9a756b2b","source":{"kind":"arxiv","id":"2409.20007","version":2},"attestation_state":"computed","paper":{"title":"DeSTA2: Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Boris Ginsburg, Chao-Han Huck Yang, Hung-yi Lee, Jagadeesh Balam, Ke-Han Lu, Szu-Wei Fu, Yu-Chiang Frank Wang, Zhehuai Chen","submitted_at":"2024-09-30T07:01:21Z","abstract_excerpt":"Recent end-to-end speech language models (SLMs) have expanded upon the capabilities of large language models (LLMs) by incorporating pre-trained speech models. However, these SLMs often undergo extensive speech instruction-tuning to bridge the gap between speech and text modalities. This requires significant annotation efforts and risks catastrophic forgetting of the original language capabilities. In this work, we present a simple yet effective automatic process for creating speech-text pair data that carefully injects speech paralinguistic understanding abilities into SLMs while preserving t"},"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":"2409.20007","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"eess.AS","submitted_at":"2024-09-30T07:01:21Z","cross_cats_sorted":["cs.CL","cs.SD"],"title_canon_sha256":"18724d28dbf2ed9717292b49a51e1cff161ed42455afe140d256eba4ccf66c5d","abstract_canon_sha256":"59e1eacf790c2a30d7f523c081898c06b6995bfc82f13ae216972e6d9028236d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:44:41.594706Z","signature_b64":"2oeuLu3JMFkCWiRwvHc54CyA10GpXeyrmR00h3T1I1B+N7BE/a1Gni+8phgqEX8XW74smL1dGuPdgQo7OheYBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6ec8da6907232f490feb28691d5b75d254bec7d1a52b94f3e9ebaefa9a756b2b","last_reissued_at":"2026-07-05T11:44:41.594212Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:44:41.594212Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DeSTA2: Developing Instruction-Following Speech Language Model Without Speech Instruction-Tuning Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Boris Ginsburg, Chao-Han Huck Yang, Hung-yi Lee, Jagadeesh Balam, Ke-Han Lu, Szu-Wei Fu, Yu-Chiang Frank Wang, Zhehuai Chen","submitted_at":"2024-09-30T07:01:21Z","abstract_excerpt":"Recent end-to-end speech language models (SLMs) have expanded upon the capabilities of large language models (LLMs) by incorporating pre-trained speech models. However, these SLMs often undergo extensive speech instruction-tuning to bridge the gap between speech and text modalities. This requires significant annotation efforts and risks catastrophic forgetting of the original language capabilities. In this work, we present a simple yet effective automatic process for creating speech-text pair data that carefully injects speech paralinguistic understanding abilities into SLMs while preserving t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2409.20007","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/2409.20007/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":"2409.20007","created_at":"2026-07-05T11:44:41.594270+00:00"},{"alias_kind":"arxiv_version","alias_value":"2409.20007v2","created_at":"2026-07-05T11:44:41.594270+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2409.20007","created_at":"2026-07-05T11:44:41.594270+00:00"},{"alias_kind":"pith_short_12","alias_value":"N3ENU2IHEMXU","created_at":"2026-07-05T11:44:41.594270+00:00"},{"alias_kind":"pith_short_16","alias_value":"N3ENU2IHEMXUSD7L","created_at":"2026-07-05T11:44:41.594270+00:00"},{"alias_kind":"pith_short_8","alias_value":"N3ENU2IH","created_at":"2026-07-05T11:44:41.594270+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.08528","citing_title":"On The Landscape of Spoken Language Models: A Comprehensive Survey","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17225","citing_title":"Can Large Audio Language Models Ignore Multilingual Distractors? An Evaluation of Their Selective Auditory Attention Capabilities","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2512.16378","citing_title":"Hearing to Translate: The Effectiveness of Speech Modality Integration into LLMs","ref_index":59,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J","json":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J.json","graph_json":"https://pith.science/api/pith-number/N3ENU2IHEMXUSD7LFBUR2W3V2J/graph.json","events_json":"https://pith.science/api/pith-number/N3ENU2IHEMXUSD7LFBUR2W3V2J/events.json","paper":"https://pith.science/paper/N3ENU2IH"},"agent_actions":{"view_html":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J","download_json":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J.json","view_paper":"https://pith.science/paper/N3ENU2IH","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2409.20007&json=true","fetch_graph":"https://pith.science/api/pith-number/N3ENU2IHEMXUSD7LFBUR2W3V2J/graph.json","fetch_events":"https://pith.science/api/pith-number/N3ENU2IHEMXUSD7LFBUR2W3V2J/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J/action/timestamp_anchor","attest_storage":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J/action/storage_attestation","attest_author":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J/action/author_attestation","sign_citation":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J/action/citation_signature","submit_replication":"https://pith.science/pith/N3ENU2IHEMXUSD7LFBUR2W3V2J/action/replication_record"}},"created_at":"2026-07-05T11:44:41.594270+00:00","updated_at":"2026-07-05T11:44:41.594270+00:00"}