{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:MHQO3MFYILRVZXJX7TX5JRYVHO","short_pith_number":"pith:MHQO3MFY","schema_version":"1.0","canonical_sha256":"61e0edb0b842e35cdd37fcefd4c7153b9c6c3e2e9888bab46dd5ba88168c3568","source":{"kind":"arxiv","id":"2605.20946","version":1},"attestation_state":"computed","paper":{"title":"Thinking-while-speaking: A Controlled, Interleaved Reasoning Method for Real-Time Speech Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Borui Jiang, Changming Xiao, Han Shu, Qiangyu Yan, Wenshuo Li, Xinghao Chen, Xuan Du","submitted_at":"2026-05-20T09:32:35Z","abstract_excerpt":"The thinking-while-speaking paradigm aims to make AI communication more human. A key challenge is maintaining fluent speech while performing deep reasoning. Our method, InterRS, tackles this by inserting reasoning steps only during natural speech generation. This requires high-quality data where reasoning and speech are precisely aligned, and the length ratio are under controlled. We introduce a novel pipeline to generate such seamlessly interleaved audio data. To train our model, we combine interleaved SFT with refined data and reinforcement learning with two new rewards: a TA-Balance Reward "},"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":"2605.20946","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-20T09:32:35Z","cross_cats_sorted":[],"title_canon_sha256":"8151f26d7cf49e1f45fb60d078780d7dad0bcabbea112abab5af4e5cf4ee85b9","abstract_canon_sha256":"7c50364661fd0fe979393e2b82c08477ac1ae3e4b07797f62fa2acda8886d470"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:29.305687Z","signature_b64":"/5YaKP2oEsByQgwgq4SxNQMMUOVR/frX2PQeF+6wU28/6cr4Y/YUX7T28U3hBxf5qF69jYFhETu1+KWB/43+Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"61e0edb0b842e35cdd37fcefd4c7153b9c6c3e2e9888bab46dd5ba88168c3568","last_reissued_at":"2026-05-21T01:05:29.305249Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:29.305249Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Thinking-while-speaking: A Controlled, Interleaved Reasoning Method for Real-Time Speech Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Borui Jiang, Changming Xiao, Han Shu, Qiangyu Yan, Wenshuo Li, Xinghao Chen, Xuan Du","submitted_at":"2026-05-20T09:32:35Z","abstract_excerpt":"The thinking-while-speaking paradigm aims to make AI communication more human. A key challenge is maintaining fluent speech while performing deep reasoning. Our method, InterRS, tackles this by inserting reasoning steps only during natural speech generation. This requires high-quality data where reasoning and speech are precisely aligned, and the length ratio are under controlled. We introduce a novel pipeline to generate such seamlessly interleaved audio data. To train our model, we combine interleaved SFT with refined data and reinforcement learning with two new rewards: a TA-Balance Reward "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.20946","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/2605.20946/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":"2605.20946","created_at":"2026-05-21T01:05:29.305302+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.20946v1","created_at":"2026-05-21T01:05:29.305302+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.20946","created_at":"2026-05-21T01:05:29.305302+00:00"},{"alias_kind":"pith_short_12","alias_value":"MHQO3MFYILRV","created_at":"2026-05-21T01:05:29.305302+00:00"},{"alias_kind":"pith_short_16","alias_value":"MHQO3MFYILRVZXJX","created_at":"2026-05-21T01:05:29.305302+00:00"},{"alias_kind":"pith_short_8","alias_value":"MHQO3MFY","created_at":"2026-05-21T01:05:29.305302+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/MHQO3MFYILRVZXJX7TX5JRYVHO","json":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO.json","graph_json":"https://pith.science/api/pith-number/MHQO3MFYILRVZXJX7TX5JRYVHO/graph.json","events_json":"https://pith.science/api/pith-number/MHQO3MFYILRVZXJX7TX5JRYVHO/events.json","paper":"https://pith.science/paper/MHQO3MFY"},"agent_actions":{"view_html":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO","download_json":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO.json","view_paper":"https://pith.science/paper/MHQO3MFY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.20946&json=true","fetch_graph":"https://pith.science/api/pith-number/MHQO3MFYILRVZXJX7TX5JRYVHO/graph.json","fetch_events":"https://pith.science/api/pith-number/MHQO3MFYILRVZXJX7TX5JRYVHO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO/action/storage_attestation","attest_author":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO/action/author_attestation","sign_citation":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO/action/citation_signature","submit_replication":"https://pith.science/pith/MHQO3MFYILRVZXJX7TX5JRYVHO/action/replication_record"}},"created_at":"2026-05-21T01:05:29.305302+00:00","updated_at":"2026-05-21T01:05:29.305302+00:00"}