MauBERT: Universal Phonetic Inductive Biases for Few-Shot Acoustic Units Discovery

TL;DR

MauBERT is a multilingual self-supervised model that incorporates articulatory features to learn language-independent phonetic representations, improving cross-lingual and casual speech understanding with minimal fine-tuning.

Contribution

It extends HuBERT with articulatory supervision across 55 languages, creating a universal phonetic model that enhances cross-lingual speech tasks.

Findings

Outperforms state-of-the-art models in ABX discriminability tests.

Effectively adapts to unseen languages and casual speech with minimal fine-tuning.

Produces more context-invariant phonetic representations.

Abstract

This paper introduces MauBERT, a multilingual extension of HuBERT that leverages articulatory features for robust cross-lingual phonetic representation learning. We continue HuBERT pre-training with supervision based on a phonetic-to-articulatory feature mapping in 55 languages. Our models learn from multilingual data to predict articulatory features or phones, resulting in language-independent representations that capture multilingual phonetic properties. Through comprehensive ABX discriminability testing, we show MauBERT models produce more context-invariant representations than state-of-the-art multilingual self-supervised learning models. Additionally, the models effectively adapt to unseen languages and casual speech with minimal self-supervised fine-tuning (10 hours of speech). This establishes an effective approach for instilling linguistic inductive biases in self-supervised speech models.