Work

PUMA: Projected Universal Multilingual ASR for Low-Resource Settings. Application to Diverse African Languages

Year: 2026

Type: conference-paper

Abstract: Multilingual ASR systems often fail to generalize to low-resource and linguistically diverse languages while remaining costly to scale.We introduce PUMA, a unified multilingual ASR model that improves low-resource performance with reduced model complexity.PUMA employs a Universal Language Projection (ULP) module that integrates a learnable language token with acoustic representations, enabling language-aware processing through shared parameters.Experiments on diverse African languages show consistent word error rate reductions over strong multilingual baselines, highlighting improved robustness and generalization.

Source: Findings of the Association for Computational Linguistics: ACL (Findings of ACL)

DOI: 10.18653/v1/2026.findings-acl.17

Authors: Ilyes Oukid, Bilal Faye, Hanane Azzag, Mustapha Lebbah, Said Yacine Boulahia

Institutions: Centre National de la Recherche Scientifique, Université Sorbonne Paris Nord, Laboratoire d'Informatique de Paris-Nord, Université de Versailles Saint-Quentin-en-Yvelines, Données et algorithmes pour une ville intelligente et durable

Language: English

Topic: Speech Recognition and Synthesis

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

OpenAlex ID: W7166835148

FWCI:

Cites:

Cited by:

Related to: 10