{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:S5QSIDQ7A2FGANPPYRR4ELKJLH","short_pith_number":"pith:S5QSIDQ7","schema_version":"1.0","canonical_sha256":"9761240e1f068a6035efc463c22d4959e8a76f97e522748ee6dea618d91decac","source":{"kind":"arxiv","id":"2606.30944","version":1},"attestation_state":"computed","paper":{"title":"Preserving Speech-to-Text LLM Capabilities in Speech-to-Speech Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Heming Wang, Heng Lu, Jian Xue, Jinyu Li, Ruchao Fan, Shuohang Wang, Xiaofei Wang, Yao Qian, Yelong Shen, Young Jin Kim, Yuxuan Hu","submitted_at":"2026-06-29T21:55:18Z","abstract_excerpt":"Strong speech-to-text (S2T) LLMs already provide robust speech perception and text reasoning, but adding speech-to-speech (S2S) output is challenging: fine-tuning the backbone can degrade the original S2T performance, while attaching a downstream talker reintroduces a serial text-to-speech bottleneck. We present PRIME-Speech, a frozen-backbone S2S conversion framework that trains only speech-generation modules. PRIME-Speech synchronizes a causal audio post-decoder with intermediate hidden states of the frozen backbone, so codec tokens are generated from the model's evolving reasoning trajector"},"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":"2606.30944","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.AS","submitted_at":"2026-06-29T21:55:18Z","cross_cats_sorted":["cs.SD"],"title_canon_sha256":"d211ddedfa358eeca1010903a8563e2fe4e8219c29f3de000047b72b60bd3719","abstract_canon_sha256":"119697e09dc2fae6ecb9f94dbbd77eea83800526696e90b0cef7f24f9c83bcb5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-01T00:17:22.263957Z","signature_b64":"9ZslD41PL0VD3avzD2Nmyt3x+OCPa7DC6/u93bSx+EtXUfocQgKmFzUQODDx5u7fJjHCqPKssfoEIwjlRPItBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9761240e1f068a6035efc463c22d4959e8a76f97e522748ee6dea618d91decac","last_reissued_at":"2026-07-01T00:17:22.263538Z","signature_status":"signed_v1","first_computed_at":"2026-07-01T00:17:22.263538Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Preserving Speech-to-Text LLM Capabilities in Speech-to-Speech Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Heming Wang, Heng Lu, Jian Xue, Jinyu Li, Ruchao Fan, Shuohang Wang, Xiaofei Wang, Yao Qian, Yelong Shen, Young Jin Kim, Yuxuan Hu","submitted_at":"2026-06-29T21:55:18Z","abstract_excerpt":"Strong speech-to-text (S2T) LLMs already provide robust speech perception and text reasoning, but adding speech-to-speech (S2S) output is challenging: fine-tuning the backbone can degrade the original S2T performance, while attaching a downstream talker reintroduces a serial text-to-speech bottleneck. We present PRIME-Speech, a frozen-backbone S2S conversion framework that trains only speech-generation modules. PRIME-Speech synchronizes a causal audio post-decoder with intermediate hidden states of the frozen backbone, so codec tokens are generated from the model's evolving reasoning trajector"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2606.30944","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/2606.30944/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":"2606.30944","created_at":"2026-07-01T00:17:22.263597+00:00"},{"alias_kind":"arxiv_version","alias_value":"2606.30944v1","created_at":"2026-07-01T00:17:22.263597+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2606.30944","created_at":"2026-07-01T00:17:22.263597+00:00"},{"alias_kind":"pith_short_12","alias_value":"S5QSIDQ7A2FG","created_at":"2026-07-01T00:17:22.263597+00:00"},{"alias_kind":"pith_short_16","alias_value":"S5QSIDQ7A2FGANPP","created_at":"2026-07-01T00:17:22.263597+00:00"},{"alias_kind":"pith_short_8","alias_value":"S5QSIDQ7","created_at":"2026-07-01T00:17:22.263597+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/S5QSIDQ7A2FGANPPYRR4ELKJLH","json":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH.json","graph_json":"https://pith.science/api/pith-number/S5QSIDQ7A2FGANPPYRR4ELKJLH/graph.json","events_json":"https://pith.science/api/pith-number/S5QSIDQ7A2FGANPPYRR4ELKJLH/events.json","paper":"https://pith.science/paper/S5QSIDQ7"},"agent_actions":{"view_html":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH","download_json":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH.json","view_paper":"https://pith.science/paper/S5QSIDQ7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2606.30944&json=true","fetch_graph":"https://pith.science/api/pith-number/S5QSIDQ7A2FGANPPYRR4ELKJLH/graph.json","fetch_events":"https://pith.science/api/pith-number/S5QSIDQ7A2FGANPPYRR4ELKJLH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH/action/storage_attestation","attest_author":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH/action/author_attestation","sign_citation":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH/action/citation_signature","submit_replication":"https://pith.science/pith/S5QSIDQ7A2FGANPPYRR4ELKJLH/action/replication_record"}},"created_at":"2026-07-01T00:17:22.263597+00:00","updated_at":"2026-07-01T00:17:22.263597+00:00"}