A Hierarchical Subspace Model for Language-Attuned Acoustic Unit Discovery

TL;DR

This paper introduces a hierarchical subspace model for acoustic unit discovery that learns language-specific units in an unsupervised manner, outperforming existing methods on multiple datasets.

Contribution

The novel hierarchical subspace approach effectively transfers phonetic knowledge across languages and improves acoustic unit discovery accuracy.

Findings

Outperforms existing acoustic unit discovery techniques.

Effective transfer of phonetic subspaces across languages.

Improved clustering and segmentation accuracy.

Abstract

In this work, we propose a hierarchical subspace model for acoustic unit discovery. In this approach, we frame the task as one of learning embeddings on a low-dimensional phonetic subspace, and simultaneously specify the subspace itself as an embedding on a hyper-subspace. We train the hyper-subspace on a set of transcribed languages and transfer it to the target language. In the target language, we infer both the language and unit embeddings in an unsupervised manner, and in so doing, we simultaneously learn a subspace of units specific to that language and the units that dwell on it. We conduct our experiments on TIMIT and two low-resource languages: Mboshi and Yoruba. Results show that our model outperforms major acoustic unit discovery techniques, both in terms of clustering quality and segmentation accuracy.