English

End-to-end streaming model for low-latency speech anonymization

Audio and Speech Processing 2024-11-04 v2 Computation and Language Machine Learning

Abstract

Speaker anonymization aims to conceal cues to speaker identity while preserving linguistic content. Current machine learning based approaches require substantial computational resources, hindering real-time streaming applications. To address these concerns, we propose a streaming model that achieves speaker anonymization with low latency. The system is trained in an end-to-end autoencoder fashion using a lightweight content encoder that extracts HuBERT-like information, a pretrained speaker encoder that extract speaker identity, and a variance encoder that injects pitch and energy information. These three disentangled representations are fed to a decoder that re-synthesizes the speech signal. We present evaluation results from two implementations of our system, a full model that achieves a latency of 230ms, and a lite version (0.1x in size) that further reduces latency to 66ms while maintaining state-of-the-art performance in naturalness, intelligibility, and privacy preservation.

Keywords

Cite

@article{arxiv.2406.09277,
  title  = {End-to-end streaming model for low-latency speech anonymization},
  author = {Waris Quamer and Ricardo Gutierrez-Osuna},
  journal= {arXiv preprint arXiv:2406.09277},
  year   = {2024}
}
R2 v1 2026-06-28T17:04:48.703Z