Differentially Private Speaker Anonymization

• URL: http://arxiv.org/abs/2202.11823v1
• Date: Wed, 23 Feb 2022 23:20:30 GMT
• Title: Differentially Private Speaker Anonymization
• Authors: Ali Shahin Shamsabadi, Brij Mohan Lal Srivastava, Aur\'elien Bellet, Nathalie Vauquier, Emmanuel Vincent, Mohamed Maouche, Marc Tommasi, Nicolas Papernot
• Abstract summary: Sharing real-world speech utterances is key to the training and deployment of voice-based services. Speaker anonymization aims to remove speaker information from a speech utterance while leaving its linguistic and prosodic attributes intact. We show that disentanglement is indeed not perfect: linguistic and prosodic attributes still contain speaker information.
• Score: 44.90119821614047
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Sharing real-world speech utterances is key to the training and deployment of voice-based services. However, it also raises privacy risks as speech contains a wealth of personal data. Speaker anonymization aims to remove speaker information from a speech utterance while leaving its linguistic and prosodic attributes intact. State-of-the-art techniques operate by disentangling the speaker information (represented via a speaker embedding) from these attributes and re-synthesizing speech based on the speaker embedding of another speaker. Prior research in the privacy community has shown that anonymization often provides brittle privacy protection, even less so any provable guarantee. In this work, we show that disentanglement is indeed not perfect: linguistic and prosodic attributes still contain speaker information. We remove speaker information from these attributes by introducing differentially private feature extractors based on an autoencoder and an automatic speech recognizer, respectively, trained using noise layers. We plug these extractors in the state-of-the-art anonymization pipeline and generate, for the first time, differentially private utterances with a provable upper bound on the speaker information they contain. We evaluate empirically the privacy and utility resulting from our differentially private speaker anonymization approach on the LibriSpeech data set. Experimental results show that the generated utterances retain very high utility for automatic speech recognition training and inference, while being much better protected against strong adversaries who leverage the full knowledge of the anonymization process to try to infer the speaker identity.

Related papers

• A Benchmark for Multi-speaker Anonymization [9.990701310620368]
  We present an attempt to provide a multi-speaker anonymization benchmark for real-world applications. A cascaded system uses speaker diarization to aggregate the speech of each speaker and speaker anonymization to conceal speaker privacy and preserve speech content. Experiments conducted on both non-overlap simulated and real-world datasets demonstrate the effectiveness of the multi-speaker anonymization system.
  arXiv  Detail & Related papers  (2024-07-08T04:48:43Z)
• Asynchronous Voice Anonymization Using Adversarial Perturbation On Speaker Embedding [46.25816642820348]
  We focus on altering the voice attributes against machine recognition while retaining human perception. A speech generation framework incorporating a speaker disentanglement mechanism is employed to generate the anonymized speech. Experiments conducted on the LibriSpeech dataset showed that the speaker attributes were obscured with their human perception preserved for 60.71% of the processed utterances.
  arXiv  Detail & Related papers  (2024-06-12T13:33:24Z)
• Evaluation of Speaker Anonymization on Emotional Speech [9.223908421919733]
  Speech data carries a range of personal information, such as the speaker's identity and emotional state. Current studies have addressed the topic of preserving speech privacy. The VoicePrivacy 2020 Challenge (VPC) is about speaker anonymization.
  arXiv  Detail & Related papers  (2023-04-15T20:50:29Z)
• Anonymizing Speech with Generative Adversarial Networks to Preserve Speaker Privacy [22.84840887071428]
  Speaker anonymization aims for hiding the identity of a speaker by changing the voice in speech recordings. This typically comes with a privacy-utility trade-off between protection of individuals and usability of the data for downstream applications. We propose to tackle this issue by generating speaker embeddings using a generative adversarial network with Wasserstein distance as cost function.
  arXiv  Detail & Related papers  (2022-10-13T13:12:42Z)
• Improving Self-Supervised Speech Representations by Disentangling Speakers [56.486084431528695]
  Self-supervised learning in speech involves training a speech representation network on a large-scale unannotated speech corpus. Disentangling speakers is very challenging, because removing the speaker information could easily result in a loss of content as well. We propose a new SSL method that can achieve speaker disentanglement without severe loss of content.
  arXiv  Detail & Related papers  (2022-04-20T04:56:14Z)
• Privacy-Preserving Speech Representation Learning using Vector Quantization [0.0]
  Speech signals contain a lot of sensitive information, such as the speaker's identity, which raises privacy concerns. This paper aims to produce an anonymous representation while preserving speech recognition performance.
  arXiv  Detail & Related papers  (2022-03-15T14:01:11Z)
• Protecting gender and identity with disentangled speech representations [49.00162808063399]
  We show that protecting gender information in speech is more effective than modelling speaker-identity information. We present a novel way to encode gender information and disentangle two sensitive biometric identifiers.
  arXiv  Detail & Related papers  (2021-04-22T13:31:41Z)
• Speaker De-identification System using Autoencoders and Adversarial Training [58.720142291102135]
  We propose a speaker de-identification system based on adversarial training and autoencoders. Experimental results show that combining adversarial learning and autoencoders increase the equal error rate of a speaker verification system.
  arXiv  Detail & Related papers  (2020-11-09T19:22:05Z)
• Improving speaker discrimination of target speech extraction with time-domain SpeakerBeam [100.95498268200777]
  SpeakerBeam exploits an adaptation utterance of the target speaker to extract his/her voice characteristics. SpeakerBeam sometimes fails when speakers have similar voice characteristics, such as in same-gender mixtures. We show experimentally that these strategies greatly improve speech extraction performance, especially for same-gender mixtures.
  arXiv  Detail & Related papers  (2020-01-23T05:36:06Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.