{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:BSVXXGJAO5KR5TYVGHOMAM426F","short_pith_number":"pith:BSVXXGJA","schema_version":"1.0","canonical_sha256":"0cab7b992077551ecf1531dcc0339af14cec46a244b5d0f42702621b01afa0ef","source":{"kind":"arxiv","id":"2406.02940","version":1},"attestation_state":"computed","paper":{"title":"Addressing Index Collapse of Large-Codebook Speech Tokenizer with Dual-Decoding Product-Quantized Variational Auto-Encoder","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Dongchao Yang, Fenglong Xie, Haohan Guo, Helen Meng, Hui Lu, Xixin Wu","submitted_at":"2024-06-05T04:54:49Z","abstract_excerpt":"VQ-VAE, as a mainstream approach of speech tokenizer, has been troubled by ``index collapse'', where only a small number of codewords are activated in large codebooks. This work proposes product-quantized (PQ) VAE with more codebooks but fewer codewords to address this problem and build large-codebook speech tokenizers. It encodes speech features into multiple VQ subspaces and composes them into codewords in a larger codebook. Besides, to utilize each VQ subspace well, we also enhance PQ-VAE via a dual-decoding training strategy with the encoding and quantized sequences. The experimental resul"},"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":"2406.02940","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SD","submitted_at":"2024-06-05T04:54:49Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"710835eafa5b7929842d1057312b8bca695ffd6c1b0bd019239e29616c10ab17","abstract_canon_sha256":"bc53b6aecdc0e1445a26e1f202026f014ff7ad70a3f44fcfa3d7c91640917425"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:27:33.023733Z","signature_b64":"2g38ljz6sDXFL8B1ddFFDgmJOGFGKGNg4zHVDHwCZZ1lBQ2BuZap8lW/F81mCN8xBbObqjm309Y8XGFU1dhtCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"0cab7b992077551ecf1531dcc0339af14cec46a244b5d0f42702621b01afa0ef","last_reissued_at":"2026-07-05T08:27:33.023263Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:27:33.023263Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Addressing Index Collapse of Large-Codebook Speech Tokenizer with Dual-Decoding Product-Quantized Variational Auto-Encoder","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Dongchao Yang, Fenglong Xie, Haohan Guo, Helen Meng, Hui Lu, Xixin Wu","submitted_at":"2024-06-05T04:54:49Z","abstract_excerpt":"VQ-VAE, as a mainstream approach of speech tokenizer, has been troubled by ``index collapse'', where only a small number of codewords are activated in large codebooks. This work proposes product-quantized (PQ) VAE with more codebooks but fewer codewords to address this problem and build large-codebook speech tokenizers. It encodes speech features into multiple VQ subspaces and composes them into codewords in a larger codebook. Besides, to utilize each VQ subspace well, we also enhance PQ-VAE via a dual-decoding training strategy with the encoding and quantized sequences. The experimental resul"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.02940","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/2406.02940/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":"2406.02940","created_at":"2026-07-05T08:27:33.023336+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.02940v1","created_at":"2026-07-05T08:27:33.023336+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.02940","created_at":"2026-07-05T08:27:33.023336+00:00"},{"alias_kind":"pith_short_12","alias_value":"BSVXXGJAO5KR","created_at":"2026-07-05T08:27:33.023336+00:00"},{"alias_kind":"pith_short_16","alias_value":"BSVXXGJAO5KR5TYV","created_at":"2026-07-05T08:27:33.023336+00:00"},{"alias_kind":"pith_short_8","alias_value":"BSVXXGJA","created_at":"2026-07-05T08:27:33.023336+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.24314","citing_title":"DS-Codec: Dual-Stage Training with Mirror-to-NonMirror Architecture Switching for Speech Codec","ref_index":24,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F","json":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F.json","graph_json":"https://pith.science/api/pith-number/BSVXXGJAO5KR5TYVGHOMAM426F/graph.json","events_json":"https://pith.science/api/pith-number/BSVXXGJAO5KR5TYVGHOMAM426F/events.json","paper":"https://pith.science/paper/BSVXXGJA"},"agent_actions":{"view_html":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F","download_json":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F.json","view_paper":"https://pith.science/paper/BSVXXGJA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.02940&json=true","fetch_graph":"https://pith.science/api/pith-number/BSVXXGJAO5KR5TYVGHOMAM426F/graph.json","fetch_events":"https://pith.science/api/pith-number/BSVXXGJAO5KR5TYVGHOMAM426F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F/action/storage_attestation","attest_author":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F/action/author_attestation","sign_citation":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F/action/citation_signature","submit_replication":"https://pith.science/pith/BSVXXGJAO5KR5TYVGHOMAM426F/action/replication_record"}},"created_at":"2026-07-05T08:27:33.023336+00:00","updated_at":"2026-07-05T08:27:33.023336+00:00"}