{"paper":{"title":"Unmixing The Crowd: Learning Persistent Speaker Representations from Mixture-Derived Multi-Speaker Embeddings","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"A model learns to predict speaker embeddings directly from noisy mixtures, enabling target speech extraction without any enrollment recording.","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Dhruv Jain, Hao-Wen Dong, Meysam Asgari, Sidharth Sidharth","submitted_at":"2026-04-03T17:46:17Z","abstract_excerpt":"We study whether persistent conversational speaker structure can be extracted directly from local overlapping speech mixtures. We propose a teacher-student framework that learns mixture-derived multi-speaker embeddings using only short overlapping segments and permutation-invariant latent supervision. Despite never being explicitly trained for speaker tracking, diarization, or conversational memory, the learned embedding space supports long-form speaker re-identification when combined with a lightweight online memory mechanism during inference. We additionally observe that the learned represen"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our model maps a noisy mixture directly to a small set of candidate speaker embeddings trained to align with a strong single-speaker speaker-embedding space via permutation-invariant teacher supervision.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That embeddings predicted from the mixture alone will align sufficiently with the single-speaker embedding space to serve as effective control signals for downstream extraction without any enrollment data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A neural model predicts a set of speaker embeddings from noisy mixtures to enable enrollment-free target speech extraction, outperforming baselines on LibriMix and generalizing to real recordings.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"A model learns to predict speaker embeddings directly from noisy mixtures, enabling target speech extraction without any enrollment recording.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"9d931b6c47b39b46cc54836b5e23e8ea88d5f77de79c43d5d3e2d4ea1b0c03a6"},"source":{"id":"2604.03219","kind":"arxiv","version":2},"verdict":{"id":"dca9a624-37ed-4531-9bc3-e2253a3f4d48","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T17:55:23.394562Z","strongest_claim":"Our model maps a noisy mixture directly to a small set of candidate speaker embeddings trained to align with a strong single-speaker speaker-embedding space via permutation-invariant teacher supervision.","one_line_summary":"A neural model predicts a set of speaker embeddings from noisy mixtures to enable enrollment-free target speech extraction, outperforming baselines on LibriMix and generalizing to real recordings.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That embeddings predicted from the mixture alone will align sufficiently with the single-speaker embedding space to serve as effective control signals for downstream extraction without any enrollment data.","pith_extraction_headline":"A model learns to predict speaker embeddings directly from noisy mixtures, enabling target speech extraction without any enrollment recording."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.03219/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":2,"snapshot_sha256":"765ffad865f890138bd07d1f47f356460c202d5f18f5bcaddf5114ae196d83cb"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}