{"paper":{"title":"HAFM: Hierarchical Autoregressive Foundation Model for Music Accompaniment Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"A hierarchical autoregressive model generates coherent instrumental accompaniments from isolated vocals using dual-rate tokenization.","cross_cats":["cs.MM"],"primary_cat":"cs.SD","authors_text":"Cheng Luo, Jianwei Cui, Jian Zhu, Jun Sun, Shihao Chen, Yubang Zhang, Yunlong Xue","submitted_at":"2026-04-10T07:27:55Z","abstract_excerpt":"Music accompaniment generation aims to automatically produce instrumental accompaniments that are rhythmically, harmonically, and timbrally coherent with a given vocal input, with broad applications in personalized music creation, arrangement assistance, and music education. Existing approaches, primarily operating in the symbolic domain or relying on single-stage audio generation frameworks, commonly suffer from insufficient high-level semantic structure modeling, limited acoustic detail reconstruction, and weak conditional controllability. To address these limitations, this paper proposes HA"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments on MUSDB18 demonstrate that HAFM achieves a Fréchet Audio Distance (FAD) of 2.08 on isolated vocal inputs, outperforming retrieval baselines and matching prior state-of-the-art systems with fewer parameters.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The dual-rate codec tokenization scheme using HuBERT at 50 Hz for vocals and EnCodec at 75 Hz for instrumentals, combined with the three-stage hierarchical autoregressive architecture, produces time-aligned and coherent accompaniments without needing additional explicit alignment or synchronization steps.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"HAFM uses a hierarchical autoregressive model with dual-rate HuBERT and EnCodec tokens to generate coherent instrumental music from vocals, achieving FAD 2.08 on MUSDB18 while matching prior systems with fewer parameters.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A hierarchical autoregressive model generates coherent instrumental accompaniments from isolated vocals using dual-rate tokenization.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"3cfdff9d5b0157a1a4694375e156ea172d81284e08f02fd5a9bc8161c39bcf16"},"source":{"id":"2604.09054","kind":"arxiv","version":3},"verdict":{"id":"2bbe5bb4-1555-4d6f-a681-853c4c16b3b9","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T17:23:06.028909Z","strongest_claim":"Experiments on MUSDB18 demonstrate that HAFM achieves a Fréchet Audio Distance (FAD) of 2.08 on isolated vocal inputs, outperforming retrieval baselines and matching prior state-of-the-art systems with fewer parameters.","one_line_summary":"HAFM uses a hierarchical autoregressive model with dual-rate HuBERT and EnCodec tokens to generate coherent instrumental music from vocals, achieving FAD 2.08 on MUSDB18 while matching prior systems with fewer parameters.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The dual-rate codec tokenization scheme using HuBERT at 50 Hz for vocals and EnCodec at 75 Hz for instrumentals, combined with the three-stage hierarchical autoregressive architecture, produces time-aligned and coherent accompaniments without needing additional explicit alignment or synchronization steps.","pith_extraction_headline":"A hierarchical autoregressive model generates coherent instrumental accompaniments from isolated vocals using dual-rate tokenization."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.09054/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"}