{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:G6Q77SYXYEWXLNW37FCB2OXNO5","short_pith_number":"pith:G6Q77SYX","schema_version":"1.0","canonical_sha256":"37a1ffcb17c12d75b6dbf9441d3aed7775a7d4c44b57193eabfcd56afff617d2","source":{"kind":"arxiv","id":"2408.16542","version":1},"attestation_state":"computed","paper":{"title":"SALSA: Speedy ASR-LLM Synchronous Aggregation","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Ashish Mittal, Darshan Prabhu, Preethi Jyothi, Sunita Sarawagi","submitted_at":"2024-08-29T14:00:57Z","abstract_excerpt":"Harnessing pre-trained LLMs to improve ASR systems, particularly for low-resource languages, is now an emerging area of research. Existing methods range from using LLMs for ASR error correction to tightly coupled systems that replace the ASR decoder with the LLM. These approaches either increase decoding time or require expensive training of the cross-attention layers. We propose SALSA, which couples the decoder layers of the ASR to the LLM decoder, while synchronously advancing both decoders. Such coupling is performed with a simple projection of the last decoder state, and is thus significan"},"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":"2408.16542","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2024-08-29T14:00:57Z","cross_cats_sorted":["cs.LG","cs.SD","eess.AS"],"title_canon_sha256":"08fe7f262d6fb6b691e69ea345e2878d4d8f3bfe62d0c980860d20ffd0bdbe8a","abstract_canon_sha256":"3e80b39853073945e0e77470c500622f5ad332135893cafe1ac1180a9685acb6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:00:42.897593Z","signature_b64":"vkEqrfCNWpfA6ZlJH1iSWqtEXbm/FePR2Ti7cKO/BrqQ0Wz0DoDZsZvWEj+4HRgy6Dtjx7ObixdhpAglWz73CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"37a1ffcb17c12d75b6dbf9441d3aed7775a7d4c44b57193eabfcd56afff617d2","last_reissued_at":"2026-07-05T09:00:42.897117Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:00:42.897117Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SALSA: Speedy ASR-LLM Synchronous Aggregation","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"","cross_cats":["cs.LG","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Ashish Mittal, Darshan Prabhu, Preethi Jyothi, Sunita Sarawagi","submitted_at":"2024-08-29T14:00:57Z","abstract_excerpt":"Harnessing pre-trained LLMs to improve ASR systems, particularly for low-resource languages, is now an emerging area of research. Existing methods range from using LLMs for ASR error correction to tightly coupled systems that replace the ASR decoder with the LLM. These approaches either increase decoding time or require expensive training of the cross-attention layers. We propose SALSA, which couples the decoder layers of the ASR to the LLM decoder, while synchronously advancing both decoders. Such coupling is performed with a simple projection of the last decoder state, and is thus significan"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.16542","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/2408.16542/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":"2408.16542","created_at":"2026-07-05T09:00:42.897173+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.16542v1","created_at":"2026-07-05T09:00:42.897173+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.16542","created_at":"2026-07-05T09:00:42.897173+00:00"},{"alias_kind":"pith_short_12","alias_value":"G6Q77SYXYEWX","created_at":"2026-07-05T09:00:42.897173+00:00"},{"alias_kind":"pith_short_16","alias_value":"G6Q77SYXYEWXLNW3","created_at":"2026-07-05T09:00:42.897173+00:00"},{"alias_kind":"pith_short_8","alias_value":"G6Q77SYX","created_at":"2026-07-05T09:00:42.897173+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2508.07285","citing_title":"Non-Intrusive Automatic Speech Recognition Refinement: A Survey","ref_index":37,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5","json":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5.json","graph_json":"https://pith.science/api/pith-number/G6Q77SYXYEWXLNW37FCB2OXNO5/graph.json","events_json":"https://pith.science/api/pith-number/G6Q77SYXYEWXLNW37FCB2OXNO5/events.json","paper":"https://pith.science/paper/G6Q77SYX"},"agent_actions":{"view_html":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5","download_json":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5.json","view_paper":"https://pith.science/paper/G6Q77SYX","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.16542&json=true","fetch_graph":"https://pith.science/api/pith-number/G6Q77SYXYEWXLNW37FCB2OXNO5/graph.json","fetch_events":"https://pith.science/api/pith-number/G6Q77SYXYEWXLNW37FCB2OXNO5/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5/action/timestamp_anchor","attest_storage":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5/action/storage_attestation","attest_author":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5/action/author_attestation","sign_citation":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5/action/citation_signature","submit_replication":"https://pith.science/pith/G6Q77SYXYEWXLNW37FCB2OXNO5/action/replication_record"}},"created_at":"2026-07-05T09:00:42.897173+00:00","updated_at":"2026-07-05T09:00:42.897173+00:00"}