Metraipolis › archive › dc_efeff2a09f3f

Jev Matches 7B Language Models for Speech-Neuroprosthesis Rescoring — arXiv:2609.33538

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.33538
License (as archived): CC0-1.0
Archived: 2026-10-07T21:53:05.730Z
SHA-256 of the archived text: 4208d86e1e42a4f35b7b50c5ac1b65c62a915e0f6c71bb389485a3f81b32cbc9
Stable URL of this page: https://metraipolis.com/documento/dc_efeff2a09f3f

How to cite

Jev Matches 7B Language Models for Speech-Neuroprosthesis Rescoring — arXiv:2609.33538 — arXiv (en), https://arxiv.org/abs/2609.33538 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_efeff2a09f3f (Metraipolis archive dc_efeff2a09f3f, sha256 4208d86e1e42a4f35b7b50c5ac1b65c62a915e0f6c71bb389485a3f81b32cbc9)

BibTeX

@misc{dcefeff2a09f3f,
  title = {Jev Matches 7B Language Models for Speech-Neuroprosthesis Rescoring — arXiv:2609.33538},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33538},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_efeff2a09f3f; sha256 4208d86e1e42a4f35b7b50c5ac1b65c62a915e0f6c71bb389485a3f81b32cbc9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33538. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A speech neuroprosthesis decodes attempted speech from brain activity and ends by rescoring the decoder's candidate sentences with a language model of several billion parameters, the only component that needs a GPU. Replacing that model with a cheaper one is hard: general language models asked to pick one sentence from a list answer from where a label sits in the list rather than from the sentence itself. We pose rescoring as a single typed decision, one call that returns a probability for every candidate, served by Jev, a hosted model trained for calibrated decisions, and combine it with the decoder's own score. On 978 held-out sentences from a participant with ALS, where the published decoder alone reaches 8.1% word error, Jev reaches 7.5% against 7.8% for both OPT-6.7b and Qwen2.5-7B; with the decoder's weight re-tuned, 6.9% against 7.2% and 7.4%. Jev is ahead in all four comparisons and at most 0.2 points behind at the 95% bound. It costs 0.07 USD per thousand sentences and needs no GPU; a dedicated GPU running a 7B model is cheaper per sentence only above 43% utilisation, far beyond what one user generates. End-to-end latency over the internet is 262 ms, of which 62 ms is spent at the provider, the same order as a 7B model on a local GPU (27 ms) but not faster.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL) ; Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2609.33538 — 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 2026-10-07T21:27:08.534Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).