Metraipolis › archive › dc_dad76ea08d5e

PTC-Bias: Phoneme-Level Temporal Competition for Bias Retrieval and Post-Decoding Correction in Speech LLMs — arXiv:2609.28727

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.28727
License (as archived)
CC0-1.0
Archived
2026-10-08T04:49:23.697Z
SHA-256 of the archived text
578eacd88961291b339800b804b6836b48c708f655ce624bc671016581846734
Stable URL of this page
https://metraipolis.com/documento/dc_dad76ea08d5e

How to cite

PTC-Bias: Phoneme-Level Temporal Competition for Bias Retrieval and Post-Decoding Correction in Speech LLMs — arXiv:2609.28727 — arXiv (en), https://arxiv.org/abs/2609.28727 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dad76ea08d5e (Metraipolis archive dc_dad76ea08d5e, sha256 578eacd88961291b339800b804b6836b48c708f655ce624bc671016581846734)
BibTeX
@misc{dcdad76ea08d5e,
  title = {PTC-Bias: Phoneme-Level Temporal Competition for Bias Retrieval and Post-Decoding Correction in Speech LLMs — arXiv:2609.28727},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.28727},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dad76ea08d5e; sha256 578eacd88961291b339800b804b6836b48c708f655ce624bc671016581846734},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.28727. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Contextual biasing improves rare-word recognition in speech large language models (SpeechLLMs), but efficiently exploiting large bias lists remains challenging. We propose PTC-Bias, a two-stage framework based on phoneme-level temporal competition. At the prefill stage, PTC Retrieval performs frame-synchronous phoneme decoding and temporal competition among candidate pronunciations, producing a compact bias-word shortlist and corresponding speech intervals. After SpeechLLM decoding, PTC Correction conducts a second local competition between the retrieved candidates and mismatched transcript spans within these intervals. Selective correction reduces near-homophone and word-segmentation errors while preserving correct transcriptions. Both stages share the same phoneme posteriors and require no additional SpeechLLM forward pass. Experiments on LibriSpeech show consistent gains across two SpeechLLMs and bias lists of up to 2000 words. With Prompt-SLAM-ASR-7B and 2000 bias words, PTC-Bias reduces B-WER by 23.4%/23.9% relative to CTC-Filter on test-clean/test-other, while keeping U-WER nearly unchanged.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-25 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Sound (cs.SD).
Pagina del lavoro: https://arxiv.org/abs/2609.28727 — 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 2026-10-08T04:27:20.019Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).