Unmixing the Crowd: Learning Mixture-to-Set Speaker Embeddings for Enrollment-Free Target Speech Extraction

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Sidharth, FNU, Asgari, Meysam, Dong, Hao-Wen, Jain, Dhruv
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910101047083008
author Sidharth, FNU
Asgari, Meysam
Dong, Hao-Wen
Jain, Dhruv
author_facet Sidharth, FNU
Asgari, Meysam
Dong, Hao-Wen
Jain, Dhruv
contents Personalized or target speech extraction (TSE) typically needs a clean enrollment -- hard to obtain in real-world crowded environments. We remove the essential need for enrollment by predicting, from the mixture itself, a small set of per-speaker embeddings that serve as the control signal for extraction. 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. On noisy LibriMix, the resulting embeddings form a structured and clusterable identity space, outperforming WavLM+K-means and separation-derived embeddings in standard clustering metrics. Conditioning these embeddings into multiple extraction back-ends consistently improves objective quality and intelligibility, and generalizes to real DNS-Challenge recordings.
format Preprint
id arxiv_https___arxiv_org_abs_2604_03219
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Unmixing the Crowd: Learning Mixture-to-Set Speaker Embeddings for Enrollment-Free Target Speech Extraction
Sidharth, FNU
Asgari, Meysam
Dong, Hao-Wen
Jain, Dhruv
Audio and Speech Processing
Sound
Personalized or target speech extraction (TSE) typically needs a clean enrollment -- hard to obtain in real-world crowded environments. We remove the essential need for enrollment by predicting, from the mixture itself, a small set of per-speaker embeddings that serve as the control signal for extraction. 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. On noisy LibriMix, the resulting embeddings form a structured and clusterable identity space, outperforming WavLM+K-means and separation-derived embeddings in standard clustering metrics. Conditioning these embeddings into multiple extraction back-ends consistently improves objective quality and intelligibility, and generalizes to real DNS-Challenge recordings.
title Unmixing the Crowd: Learning Mixture-to-Set Speaker Embeddings for Enrollment-Free Target Speech Extraction
topic Audio and Speech Processing
Sound
url https://arxiv.org/abs/2604.03219