{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2021:5H554CIWO7QQGNWK6DD7M5S7Q7","short_pith_number":"pith:5H554CIW","schema_version":"1.0","canonical_sha256":"e9fbde091677e10336caf0c7f6765f87fae4e2cb720ef9b04aed9ef9a4546b0d","source":{"kind":"arxiv","id":"2103.15060","version":3},"attestation_state":"computed","paper":{"title":"PnG BERT: Augmented BERT on Phonemes and Graphemes for Neural TTS","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Heiga Zen, Jonathan Shen, Ye Jia, Yonghui Wu, Yu Zhang","submitted_at":"2021-03-28T06:24:00Z","abstract_excerpt":"This paper introduces PnG BERT, a new encoder model for neural TTS. This model is augmented from the original BERT model, by taking both phoneme and grapheme representations of text as input, as well as the word-level alignment between them. It can be pre-trained on a large text corpus in a self-supervised manner, and fine-tuned in a TTS task. Experimental results show that a neural TTS model using a pre-trained PnG BERT as its encoder yields more natural prosody and more accurate pronunciation than a baseline model using only phoneme input with no pre-training. Subjective side-by-side prefere"},"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":"2103.15060","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2021-03-28T06:24:00Z","cross_cats_sorted":["cs.SD","eess.AS"],"title_canon_sha256":"c515173c4a7d4208163a90dc270a6f83d9e538ba88bba2f9f8954117e5ccfdd1","abstract_canon_sha256":"7da762b595981ed96aa42e37ee06ab40af33c8da5f8cb155df98e17b2f961a75"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T02:46:45.186013Z","signature_b64":"abVZlOQoXYbRjIyt3UrEmfMsi8NWhH1kAkCP0eEdbf3gIxgtL7yWj5CgQMp3R2dOs7Yr+ZaADqHno4DZbyTdCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e9fbde091677e10336caf0c7f6765f87fae4e2cb720ef9b04aed9ef9a4546b0d","last_reissued_at":"2026-07-05T02:46:45.185393Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T02:46:45.185393Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"PnG BERT: Augmented BERT on Phonemes and Graphemes for Neural TTS","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Heiga Zen, Jonathan Shen, Ye Jia, Yonghui Wu, Yu Zhang","submitted_at":"2021-03-28T06:24:00Z","abstract_excerpt":"This paper introduces PnG BERT, a new encoder model for neural TTS. This model is augmented from the original BERT model, by taking both phoneme and grapheme representations of text as input, as well as the word-level alignment between them. It can be pre-trained on a large text corpus in a self-supervised manner, and fine-tuned in a TTS task. Experimental results show that a neural TTS model using a pre-trained PnG BERT as its encoder yields more natural prosody and more accurate pronunciation than a baseline model using only phoneme input with no pre-training. Subjective side-by-side prefere"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2103.15060","kind":"arxiv","version":3},"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/2103.15060/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":"2103.15060","created_at":"2026-07-05T02:46:45.185499+00:00"},{"alias_kind":"arxiv_version","alias_value":"2103.15060v3","created_at":"2026-07-05T02:46:45.185499+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2103.15060","created_at":"2026-07-05T02:46:45.185499+00:00"},{"alias_kind":"pith_short_12","alias_value":"5H554CIWO7QQ","created_at":"2026-07-05T02:46:45.185499+00:00"},{"alias_kind":"pith_short_16","alias_value":"5H554CIWO7QQGNWK","created_at":"2026-07-05T02:46:45.185499+00:00"},{"alias_kind":"pith_short_8","alias_value":"5H554CIW","created_at":"2026-07-05T02:46:45.185499+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.10980","citing_title":"Magnetic field configuration favoring X-class solar flares: violation of Joy's and Hale's laws","ref_index":47,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7","json":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7.json","graph_json":"https://pith.science/api/pith-number/5H554CIWO7QQGNWK6DD7M5S7Q7/graph.json","events_json":"https://pith.science/api/pith-number/5H554CIWO7QQGNWK6DD7M5S7Q7/events.json","paper":"https://pith.science/paper/5H554CIW"},"agent_actions":{"view_html":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7","download_json":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7.json","view_paper":"https://pith.science/paper/5H554CIW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2103.15060&json=true","fetch_graph":"https://pith.science/api/pith-number/5H554CIWO7QQGNWK6DD7M5S7Q7/graph.json","fetch_events":"https://pith.science/api/pith-number/5H554CIWO7QQGNWK6DD7M5S7Q7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7/action/storage_attestation","attest_author":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7/action/author_attestation","sign_citation":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7/action/citation_signature","submit_replication":"https://pith.science/pith/5H554CIWO7QQGNWK6DD7M5S7Q7/action/replication_record"}},"created_at":"2026-07-05T02:46:45.185499+00:00","updated_at":"2026-07-05T02:46:45.185499+00:00"}