{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:64XXGIGJGNDD5GSW4RQFAKNLK3","short_pith_number":"pith:64XXGIGJ","schema_version":"1.0","canonical_sha256":"f72f7320c933463e9a56e4605029ab56c4a44d84ddc63c5a07f4279608b0ae6f","source":{"kind":"arxiv","id":"2410.22179","version":2},"attestation_state":"computed","paper":{"title":"Robust and Unbounded Length Generalization in Autoregressive Transformer-Based Text-to-Speech","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Daisy Stanton, David Kao, Eric Battenberg, Julian Salazar, Matt Shannon, RJ Skerry-Ryan, Soroosh Mariooryad","submitted_at":"2024-10-29T16:17:01Z","abstract_excerpt":"Autoregressive (AR) Transformer-based sequence models are known to have difficulty generalizing to sequences longer than those seen during training. When applied to text-to-speech (TTS), these models tend to drop or repeat words or produce erratic output, especially for longer utterances. In this paper, we introduce enhancements aimed at AR Transformer-based encoder-decoder TTS systems that address these robustness and length generalization issues. Our approach uses an alignment mechanism to provide cross-attention operations with relative location information. The associated alignment positio"},"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":"2410.22179","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-29T16:17:01Z","cross_cats_sorted":["cs.LG","cs.SD","eess.AS"],"title_canon_sha256":"1287545e0148847c79fbf7351fbfa0af04ac74f01559b582773e85917a183049","abstract_canon_sha256":"413e88a60dc24366f1f39289692a0259f38e1537af3a05b916a923c4eec55933"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:29:25.736002Z","signature_b64":"pgkHNCSzdFt5TWtQotEXCKfBmy91vCL6RRlgYKlUgHjIrVIG4GUCn94bArZgU3HRQhXLWqr0HnTYD+KTrHqDBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f72f7320c933463e9a56e4605029ab56c4a44d84ddc63c5a07f4279608b0ae6f","last_reissued_at":"2026-07-05T10:29:25.735366Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:29:25.735366Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Robust and Unbounded Length Generalization in Autoregressive Transformer-Based Text-to-Speech","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Daisy Stanton, David Kao, Eric Battenberg, Julian Salazar, Matt Shannon, RJ Skerry-Ryan, Soroosh Mariooryad","submitted_at":"2024-10-29T16:17:01Z","abstract_excerpt":"Autoregressive (AR) Transformer-based sequence models are known to have difficulty generalizing to sequences longer than those seen during training. When applied to text-to-speech (TTS), these models tend to drop or repeat words or produce erratic output, especially for longer utterances. In this paper, we introduce enhancements aimed at AR Transformer-based encoder-decoder TTS systems that address these robustness and length generalization issues. Our approach uses an alignment mechanism to provide cross-attention operations with relative location information. The associated alignment positio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.22179","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/2410.22179/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":"2410.22179","created_at":"2026-07-05T10:29:25.735463+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.22179v2","created_at":"2026-07-05T10:29:25.735463+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.22179","created_at":"2026-07-05T10:29:25.735463+00:00"},{"alias_kind":"pith_short_12","alias_value":"64XXGIGJGNDD","created_at":"2026-07-05T10:29:25.735463+00:00"},{"alias_kind":"pith_short_16","alias_value":"64XXGIGJGNDD5GSW","created_at":"2026-07-05T10:29:25.735463+00:00"},{"alias_kind":"pith_short_8","alias_value":"64XXGIGJ","created_at":"2026-07-05T10:29:25.735463+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.19462","citing_title":"VoiceStar: Robust Zero-Shot Autoregressive TTS with Duration Control and Extrapolation","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3","json":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3.json","graph_json":"https://pith.science/api/pith-number/64XXGIGJGNDD5GSW4RQFAKNLK3/graph.json","events_json":"https://pith.science/api/pith-number/64XXGIGJGNDD5GSW4RQFAKNLK3/events.json","paper":"https://pith.science/paper/64XXGIGJ"},"agent_actions":{"view_html":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3","download_json":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3.json","view_paper":"https://pith.science/paper/64XXGIGJ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.22179&json=true","fetch_graph":"https://pith.science/api/pith-number/64XXGIGJGNDD5GSW4RQFAKNLK3/graph.json","fetch_events":"https://pith.science/api/pith-number/64XXGIGJGNDD5GSW4RQFAKNLK3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3/action/storage_attestation","attest_author":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3/action/author_attestation","sign_citation":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3/action/citation_signature","submit_replication":"https://pith.science/pith/64XXGIGJGNDD5GSW4RQFAKNLK3/action/replication_record"}},"created_at":"2026-07-05T10:29:25.735463+00:00","updated_at":"2026-07-05T10:29:25.735463+00:00"}