X-vector anonymization using autoencoders and adversarial training for preserving speech privacy.
Juan M. Perero-CodoseroFernando M. Espinoza-CuadrosLuis A. Hernández GómezPublished in: Comput. Speech Lang. (2022)
Keyphrases
- privacy preserving
- privacy protection
- differential privacy
- privacy preservation
- preserving privacy
- information loss
- data anonymization
- hearing impaired
- data publishing
- data privacy
- anonymized data
- privacy preserving data publishing
- privacy preserving data mining
- training set
- personal information
- privacy guarantees
- denoising
- feedforward neural networks
- training algorithm
- supervised learning
- speech recognition
- private data
- privacy issues
- sensitive information
- spoken language
- social network data
- output layer
- training process
- neural network
- privacy concerns
- training samples
- social network analysis
- hidden markov models
- privacy requirements
- cloud computing
- speech signal
- probabilistic model
- statistical databases
- private information