MauBERT-feat

MauBERT-feat is a speech model adapting a pre-trained HuBERT base model to predict frame-level articulatory features on a subset of 788 hours and 55 languages of Common Voice 16.1 leveraging the phone annotations from VoxCommunis. It was introduced by the MauBERT paper at ACL 2026.

You can load it with maubert:

from maubert import MauBERTFeat

model = MauBERTFeat.from_pretrained("coml/maubert-feat")

Check out the maubert repository if you are interested in further retraining or fine-tuning, and the corresponding paper for our results.

Citing

@inproceedings{ortiztandazo-etal-2026-maubert,
    title = "{M}au{BERT}: Universal Phonetic Inductive Biases for Few-Shot Acoustic Units Discovery",
    author = "Ortiz Tandazo, Angelo  and
      Khentout, Manel  and
      Benchekroun, Youssef  and
      Hueber, Thomas  and
      Dupoux, Emmanuel",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.acl-long.24/",
    doi = "10.18653/v1/2026.acl-long.24",
    pages = "568--585",
    ISBN = "979-8-89176-390-6",
}