HuPER: A Human-Inspired Framework for Phonetic Perception

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Guo, Chenxu, Lian, Jiachen, Liu, Yisi, Huang, Baihe, Narayanan, Shriyaa, Cho, Cheol Jun, Anumanchipalli, Gopala
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915768471388160
author Guo, Chenxu
Lian, Jiachen
Liu, Yisi
Huang, Baihe
Narayanan, Shriyaa
Cho, Cheol Jun
Anumanchipalli, Gopala
author_facet Guo, Chenxu
Lian, Jiachen
Liu, Yisi
Huang, Baihe
Narayanan, Shriyaa
Cho, Cheol Jun
Anumanchipalli, Gopala
contents We propose HuPER, a human-inspired framework that models phonetic perception as adaptive inference over acoustic-phonetics evidence and linguistic knowledge. With only 100 hours of training data, HuPER achieves state-of-the-art phonetic error rates on five English benchmarks and strong zero-shot transfer to 95 unseen languages. HuPER is also the first framework to enable adaptive, multi-path phonetic perception under diverse acoustic conditions. All training data, models, and code are open-sourced. Code and demo avaliable at https://github.com/HuPER29/HuPER.
format Preprint
id arxiv_https___arxiv_org_abs_2602_01634
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HuPER: A Human-Inspired Framework for Phonetic Perception
Guo, Chenxu
Lian, Jiachen
Liu, Yisi
Huang, Baihe
Narayanan, Shriyaa
Cho, Cheol Jun
Anumanchipalli, Gopala
Audio and Speech Processing
Artificial Intelligence
We propose HuPER, a human-inspired framework that models phonetic perception as adaptive inference over acoustic-phonetics evidence and linguistic knowledge. With only 100 hours of training data, HuPER achieves state-of-the-art phonetic error rates on five English benchmarks and strong zero-shot transfer to 95 unseen languages. HuPER is also the first framework to enable adaptive, multi-path phonetic perception under diverse acoustic conditions. All training data, models, and code are open-sourced. Code and demo avaliable at https://github.com/HuPER29/HuPER.
title HuPER: A Human-Inspired Framework for Phonetic Perception
topic Audio and Speech Processing
Artificial Intelligence
url https://arxiv.org/abs/2602.01634