Metraipolis › archive › dc_03493742eebc

Neural Representations, Natural Connections: What Transfers From Human Speech Foundation Models to Animal Vocalizations? — arXiv:2610.07107

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.07107
License (as archived)
CC0-1.0
Archived
2026-10-07T21:22:47.141Z
SHA-256 of the archived text
a91bb233dedb58b874f189549acd4eed558fc01557abcd973852f1620253f110
Stable URL of this page
https://metraipolis.com/documento/dc_03493742eebc

How to cite

Neural Representations, Natural Connections: What Transfers From Human Speech Foundation Models to Animal Vocalizations? — arXiv:2610.07107 — arXiv (en), https://arxiv.org/abs/2610.07107 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_03493742eebc (Metraipolis archive dc_03493742eebc, sha256 a91bb233dedb58b874f189549acd4eed558fc01557abcd973852f1620253f110)
BibTeX
@misc{dc03493742eebc,
  title = {Neural Representations, Natural Connections: What Transfers From Human Speech Foundation Models to Animal Vocalizations? — arXiv:2610.07107},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07107},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_03493742eebc; sha256 a91bb233dedb58b874f189549acd4eed558fc01557abcd973852f1620253f110},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07107. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speech-pretrained models have shown promise in animal bioacoustics, but the factors governing their cross-species transfer remain poorly understood. We evaluate 15 frozen encoders spanning monolingual and multilingual speech, speaker verification, animal bioacoustics, and general audio on caller identification across four species and call-type classification across three. Differences between the strongest speech- and animal-pretrained representations range from $-0.027$ to $+0.119$ UAR; speech is significantly better in three of seven settings ($p<0.01$) and not significantly different in the remaining four. Controlled comparisons show no systematic advantage from increased language coverage or animal-domain pretraining, while frozen speaker-verification embeddings transfer poorly. The results also show that transfer is strongly layer-dependent: raw-waveform models peak early, whereas patch-spectrogram models peak deeper.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv eess.AS, nuovi lavori del giorno).
Categorie arXiv: Sound (cs.SD) ; Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2610.07107 — 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 2026-10-07T20:54:04.262Z dalla pagina https://arxiv.org/list/eess.AS/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).