Metraipolis › archive › dc_3fe3308f504b

Asymmetric Classifier-Free Guidance for Target-Speaker ASR — arXiv:2609.30476

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.30476
License (as archived)
CC0-1.0
Archived
2026-10-08T02:04:12.733Z
SHA-256 of the archived text
178cbfe03f37f1d4bb740acfd908236652bbe8f36e796ccdfe2b4a9de0e88aff
Stable URL of this page
https://metraipolis.com/documento/dc_3fe3308f504b

How to cite

Asymmetric Classifier-Free Guidance for Target-Speaker ASR — arXiv:2609.30476 — arXiv (en), https://arxiv.org/abs/2609.30476 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3fe3308f504b (Metraipolis archive dc_3fe3308f504b, sha256 178cbfe03f37f1d4bb740acfd908236652bbe8f36e796ccdfe2b4a9de0e88aff)
BibTeX
@misc{dc3fe3308f504b,
  title = {Asymmetric Classifier-Free Guidance for Target-Speaker ASR — arXiv:2609.30476},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.30476},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3fe3308f504b; sha256 178cbfe03f37f1d4bb740acfd908236652bbe8f36e796ccdfe2b4a9de0e88aff},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.30476. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Target-speaker automatic speech recognition (TS-ASR) must identify and transcribe a desired speaker under varying overlap and noise conditions. These changes alter the acoustic evidence for the target speaker in the speech mixture, motivating inference-time calibration of speaker conditioning. We introduce asymmetric classifier-free guidance (CFG) for TS-ASR using Whisper: the speaker-conditioned branch predicts the target transcript, while the speaker-unconditioned branch predicts serialized multi-speaker transcripts. CFG adjusts the contribution of speaker conditioning during decoding through a single guidance scale. We select a global guidance scale on target-domain development data and train a lightweight encoder-based predictor to adjust it for each utterance, keeping the recognition model fixed. Under domain shifts, our full system achieves relative word error rate (WER) reductions of up to 21.8% over the condition-only baseline, and 5.6% over standard conditional decoding of the same CFG-trained model. Oracle analysis shows that substantially larger WER reductions are possible through utterance-level scale selection and identifies how beneficial adjustments vary with domain shifts.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-28 in cs.CL).
Categorie arXiv: Audio and Speech Processing (eess.AS) ; Computation and Language (cs.CL); Sound (cs.SD).
Pagina del lavoro: https://arxiv.org/abs/2609.30476 — 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 2026-10-08T01:38:49.473Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-28?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).