{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:ODNUGYYPYZ5Q6RNN522UH2WXBL","short_pith_number":"pith:ODNUGYYP","schema_version":"1.0","canonical_sha256":"70db43630fc67b0f45adeeb543ead70aecbaf84c5197e624f085d100397d5795","source":{"kind":"arxiv","id":"2504.15509","version":1},"attestation_state":"computed","paper":{"title":"SimulS2S-LLM: Unlocking Simultaneous Inference of Speech LLMs for Speech-to-Speech Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Keqi Deng, Philip C. Woodland, Wenxi Chen, Xie Chen","submitted_at":"2025-04-22T01:05:32Z","abstract_excerpt":"Simultaneous speech translation (SST) outputs translations in parallel with streaming speech input, balancing translation quality and latency. While large language models (LLMs) have been extended to handle the speech modality, streaming remains challenging as speech is prepended as a prompt for the entire generation process. To unlock LLM streaming capability, this paper proposes SimulS2S-LLM, which trains speech LLMs offline and employs a test-time policy to guide simultaneous inference. SimulS2S-LLM alleviates the mismatch between training and inference by extracting boundary-aware speech p"},"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":"2504.15509","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-22T01:05:32Z","cross_cats_sorted":["cs.SD","eess.AS"],"title_canon_sha256":"8e20f82599a3946c32a9dc8f26e6457345534fd83d4443fe34136eabb370dac1","abstract_canon_sha256":"e144a3079f73ec91fd1c369036ca86d6a43192c69525d94deb0e636f05cf8d94"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:52:25.728742Z","signature_b64":"Tj2ubDOFeNQsMOKFmyhG/L3SsV469rRJPoAc3kUBwRNkx1bMow6cDNHCtFbIk0FdB18bRCuShTijPxEiRo0ZDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"70db43630fc67b0f45adeeb543ead70aecbaf84c5197e624f085d100397d5795","last_reissued_at":"2026-07-05T10:52:25.728109Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:52:25.728109Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SimulS2S-LLM: Unlocking Simultaneous Inference of Speech LLMs for Speech-to-Speech Translation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Keqi Deng, Philip C. Woodland, Wenxi Chen, Xie Chen","submitted_at":"2025-04-22T01:05:32Z","abstract_excerpt":"Simultaneous speech translation (SST) outputs translations in parallel with streaming speech input, balancing translation quality and latency. While large language models (LLMs) have been extended to handle the speech modality, streaming remains challenging as speech is prepended as a prompt for the entire generation process. To unlock LLM streaming capability, this paper proposes SimulS2S-LLM, which trains speech LLMs offline and employs a test-time policy to guide simultaneous inference. SimulS2S-LLM alleviates the mismatch between training and inference by extracting boundary-aware speech p"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.15509","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/2504.15509/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":"2504.15509","created_at":"2026-07-05T10:52:25.728176+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.15509v1","created_at":"2026-07-05T10:52:25.728176+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.15509","created_at":"2026-07-05T10:52:25.728176+00:00"},{"alias_kind":"pith_short_12","alias_value":"ODNUGYYPYZ5Q","created_at":"2026-07-05T10:52:25.728176+00:00"},{"alias_kind":"pith_short_16","alias_value":"ODNUGYYPYZ5Q6RNN","created_at":"2026-07-05T10:52:25.728176+00:00"},{"alias_kind":"pith_short_8","alias_value":"ODNUGYYP","created_at":"2026-07-05T10:52:25.728176+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.08486","citing_title":"TRADE: Transducer-Augmented Decoder for Speech LLM","ref_index":8,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL","json":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL.json","graph_json":"https://pith.science/api/pith-number/ODNUGYYPYZ5Q6RNN522UH2WXBL/graph.json","events_json":"https://pith.science/api/pith-number/ODNUGYYPYZ5Q6RNN522UH2WXBL/events.json","paper":"https://pith.science/paper/ODNUGYYP"},"agent_actions":{"view_html":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL","download_json":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL.json","view_paper":"https://pith.science/paper/ODNUGYYP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.15509&json=true","fetch_graph":"https://pith.science/api/pith-number/ODNUGYYPYZ5Q6RNN522UH2WXBL/graph.json","fetch_events":"https://pith.science/api/pith-number/ODNUGYYPYZ5Q6RNN522UH2WXBL/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL/action/storage_attestation","attest_author":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL/action/author_attestation","sign_citation":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL/action/citation_signature","submit_replication":"https://pith.science/pith/ODNUGYYPYZ5Q6RNN522UH2WXBL/action/replication_record"}},"created_at":"2026-07-05T10:52:25.728176+00:00","updated_at":"2026-07-05T10:52:25.728176+00:00"}