Contrastive prediction strategies for unsupervised segmentation and categorization of phonemes and words

TL;DR

This paper explores contrastive predictive coding methods for unsupervised phoneme and word segmentation and categorization, identifying a trade-off caused by context networks and proposing a multi-level model to improve both tasks.

Contribution

The paper introduces multi-level ACPC (mACPC), a novel variation of CPC that enhances phoneme categorization and achieves state-of-the-art word segmentation performance.

Findings

mACPC outperforms previous models in categorization metrics

mACPC achieves state-of-the-art results in word segmentation

Using multi-level modeling reduces the trade-off between segmentation and categorization

Abstract

We investigate the performance on phoneme categorization and phoneme and word segmentation of several self-supervised learning (SSL) methods based on Contrastive Predictive Coding (CPC). Our experiments show that with the existing algorithms there is a trade off between categorization and segmentation performance. We investigate the source of this conflict and conclude that the use of context building networks, albeit necessary for superior performance on categorization tasks, harms segmentation performance by causing a temporal shift on the learned representations. Aiming to bridge this gap, we take inspiration from the leading approach on segmentation, which simultaneously models the speech signal at the frame and phoneme level, and incorporate multi-level modelling into Aligned CPC (ACPC), a variation of CPC which exhibits the best performance on categorization tasks. Our multi-level ACPC (mACPC) improves in all categorization metrics and achieves state-of-the-art performance in word segmentation.