Exploring How Generative Adversarial Networks Learn Phonological Representations
Jingyi Chen, Micha Elsner
Abstract
This paper explores how Generative Adversarial Networks (GANs) learn representations of phonological phenomena. We analyze how GANs encode contrastive and non-contrastive nasality in French and English vowels by applying the ciwGAN architecture (Beguš, 2021a). Beguš claims that ciwGAN encodes linguistically meaningful representations with categorical variables in its latent space and manipulating the latent variables shows an almost one to one corresponding control of the phonological features in ciwGAN's generated outputs. However, our results show an interactive effect of latent variables on the features in the generated outputs, which suggests the learned representations in neural networks are different from the phonological representations proposed by linguists. On the other hand, ci-wGAN is able to distinguish contrastive and noncontrastive features in English and French by encoding them differently. Comparing the performance of GANs learning from different languages results in a better understanding of what language specific features contribute to developing language specific phonological representations. We also discuss the role of training data frequencies in phonological feature learning.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on2
- wav2vec 2.0: A Framework for Self-Supervised Learning of Speech RepresentationsAlexei Baevski, Yuhao Zhou, Abdelrahman Mohamed, Michael AuliNeurIPS 2020 · 9,451 citations
- vq-wav2vec: Self-Supervised Learning of Discrete Speech RepresentationsAlexei Baevski, Steffen Schneider, Michael AuliICLR 2020 · 730 citations
Related papers
- Counter-Contrastive Learning for Language GANsYekun Chai, Haidong Zhang, Qiyue Yin, Junge ZhangEMNLP Findings 2021 · 4 citations
- Training GANs with Stronger Augmentations via Contrastive DiscriminatorJongheon Jeong, Jinwoo ShinICLR 2021 · 68 citations
- Interpreting the Latent Space of GANs for Semantic Face EditingYujun Shen, Jinjin Gu, Xiaoou Tang, Bolei ZhouCVPR 2020
- Lip to Speech Synthesis with Visual Context Attentional GANMinsu Kim, Joanna Hong, Yong Man RoNeurIPS 2021 · 76 citations
- BigVGAN: A Universal Neural Vocoder with Large-Scale TrainingSang-gil Lee, Wei Ping, Boris Ginsburg, Bryan Catanzaro et al.ICLR 2023 · 46 citations
