Generating gender-ambiguous voices for privacy-preserving speech recognition
This addresses privacy concerns for individuals using speech recognition services by concealing sensitive attributes like gender, though it is incremental as it builds on prior privacy-preserving techniques.
The paper tackles the problem of preventing gender and identity inference from voice data in speech recognition by proposing GenGAN, a generative adversarial network that synthesizes gender-ambiguous voices, showing it improves the privacy-utility trade-off compared to existing methods.
Our voice encodes a uniquely identifiable pattern which can be used to infer private attributes, such as gender or identity, that an individual might wish not to reveal when using a speech recognition service. To prevent attribute inference attacks alongside speech recognition tasks, we present a generative adversarial network, GenGAN, that synthesises voices that conceal the gender or identity of a speaker. The proposed network includes a generator with a U-Net architecture that learns to fool a discriminator. We condition the generator only on gender information and use an adversarial loss between signal distortion and privacy preservation. We show that GenGAN improves the trade-off between privacy and utility compared to privacy-preserving representation learning methods that consider gender information as a sensitive attribute to protect.