Metraipolis › archive › dc_e888b5ce4d37

Unified Target-Speaker ASR with Text and Enrollment Speech Cues — arXiv:2609.33853

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33853
License (as archived)
CC0-1.0
Archived
2026-10-08T01:36:52.206Z
SHA-256 of the archived text
031878c9c3a3bc5ba02b3eb6d80cccd8d7babc8433135b286226462c0920b54a
Stable URL of this page
https://metraipolis.com/documento/dc_e888b5ce4d37

How to cite

Unified Target-Speaker ASR with Text and Enrollment Speech Cues — arXiv:2609.33853 — arXiv (en), https://arxiv.org/abs/2609.33853 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_e888b5ce4d37 (Metraipolis archive dc_e888b5ce4d37, sha256 031878c9c3a3bc5ba02b3eb6d80cccd8d7babc8433135b286226462c0920b54a)
BibTeX
@misc{dce888b5ce4d37,
  title = {Unified Target-Speaker ASR with Text and Enrollment Speech Cues — arXiv:2609.33853},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33853},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_e888b5ce4d37; sha256 031878c9c3a3bc5ba02b3eb6d80cccd8d7babc8433135b286226462c0920b54a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33853. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Target-speaker automatic speech recognition (TS-ASR) aims to recognize a designated speaker while suppressing interfering speech in multi-talker environments. Conventional TS-ASR typically relies on an enrollment utterance, whereas text-guided methods use known lexical content, such as a wake word, to identify the target speaker from the observed mixture. These two cues provide complementary information but are usually studied separately. We propose a Unified Dual-Cue TS-ASR framework that supports text cues, enrollment speech, or both within a single model. Text cues interact with the mixture representation to extract target-speaker information conditioned on known lexical content, while an independent enrollment utterance provides complementary speaker information. Cross-attention cue-conditioning modules are integrated into shared Conformer blocks, and negative-cue sampling provides cue-validity supervision during dual-cue training. Experiments on 30,000 two-speaker mixtures across five recording/domain conditions and four oracle text-cue lengths show that, with five-character text cues, the concatenated dual-cue method achieves 8.80% CER, compared with 17.32% for text-only and 29.06% for enrollment-only inference. It also outperforms parallel dual-cue fusion (9.49% CER) and yields lower dual-cue CER across all five evaluation subsets. These results demonstrate the benefit of jointly exploiting complementary lexical and speaker information for target-speaker ASR.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-29 in eess.SP).
Categorie arXiv: Signal Processing (eess.SP) ; Sound (cs.SD); Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2609.33853 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T01:11:02.818Z dalla pagina https://arxiv.org/catchup/eess.SP/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).