Voiceprint verification has been widely adopted in daily applications due to its convenience and efficiency. However, the latent representations extracted during speech verification often encode sensitive personal information, such as emotion and gender, which raises significant privacy concerns. Existing methods mainly focus on protecting predefined sensitive attributes, but in many cases, the attributes that attackers are interested in are unknown to defenders. To address this limitation, we propose the Speech Information Bottleneck Network (SIBNet), a framework for representation level privacy enhancement based on information theoretic approaches. SIBNet differs from existing speech attribute privacy mechanisms in three key respects: (1) We formulate biometric privacy protection as an Information Bottleneck (IB) optimization problem that balances identity utility and sensitive attribute suppression. (2) SIBNet leverages a variational distribution parameterized by neural networks to approximate mutual information, enabling the derivation of lower bounds for effective model optimization. (3) An additive angular margin penalty is introduced into the speaker classification loss to promote discriminative embeddings and preserve identity utility. Experiments on public speech datasets demonstrate that SIBNet significantly outperforms state of the art methods in both biometric recognition accuracy and suppression of sensitive soft biometric attributes.

Consumer electronic devices with voice assistants are becoming increasingly popular in modern intelligent home. Nevertheless, directly uploading unprocessed speech data, which may contain sensitive attributes, to a cloud server poses a significant risk to user privacy. To address this privacy issue, this paper proposes a privacy-enhancing model to protect speech emotions based on generative adversarial networks (PSEGAN). The model aims to prevent the inference of emotional attributes while maintaining the accuracy and utility of speech features. PSEGAN benefits from three modules: (1) A pre-trained speaker matcher imposes generative constraints on the model during the training phase to ensure that the generated speech retains the essential information needed for speaker recognition. (2) Attribute adversarial networks can generate perturbed speech that transforms emotional attributes while preserving the utility of the speech. (3) Gated Recurrent Networks (GRN) can handle the long-short term dependencies of speech signals. PSEGAN model solves the problem of utility loss in traditional speech privacy preservation methods based on generative adversarial networks (GAN). Experimental results show that on the RAVDESS dataset, PSEGAN reduces emotion recognition accuracy by 80.7%, while speaker recognition accuracy only decreases by 1.1%. These findings demonstrate that PSEGAN effectively mitigates the leakage of emotional attributes while maintaining high utility.