English

Learning Robust Latent Representations for Controllable Speech Synthesis

Sound 2021-05-11 v1 Computation and Language Machine Learning Audio and Speech Processing

Abstract

State-of-the-art Variational Auto-Encoders (VAEs) for learning disentangled latent representations give impressive results in discovering features like pitch, pause duration, and accent in speech data, leading to highly controllable text-to-speech (TTS) synthesis. However, these LSTM-based VAEs fail to learn latent clusters of speaker attributes when trained on either limited or noisy datasets. Further, different latent variables start encoding the same features, limiting the control and expressiveness during speech synthesis. To resolve these issues, we propose RTI-VAE (Reordered Transformer with Information reduction VAE) where we minimize the mutual information between different latent variables and devise a modified Transformer architecture with layer reordering to learn controllable latent representations in speech data. We show that RTI-VAE reduces the cluster overlap of speaker attributes by at least 30\% over LSTM-VAE and by at least 7\% over vanilla Transformer-VAE.

Keywords

Cite

@article{arxiv.2105.04458,
  title  = {Learning Robust Latent Representations for Controllable Speech Synthesis},
  author = {Shakti Kumar and Jithin Pradeep and Hussain Zaidi},
  journal= {arXiv preprint arXiv:2105.04458},
  year   = {2021}
}

Comments

Accepted in ACL2021 Findings

R2 v1 2026-06-24T01:57:10.616Z