The invention
DeepRawNet: empowering deepfake audio detection through dynamic enhancements
W: Hardware – Software – Cybersecurity – Blockchain – Internet of things (IoT)
Informations
- Stand number
- K96
- Exhibition class
- W: Hardware – Software – Cybersecurity – Blockchain – Internet of things (IoT)
- Technical description
- DeepRawNet is an advanced audio spoofing detection framework that improves RawNet2 by refining Sinc filters, upgrading LeakyReLU to PReLU, and replacing convolution layers with transpose convolutions in residual blocks, enhancing detection accuracy.
- Simplified description
- DeepRawNet is an improved audio spoofing detection system that builds on RawNet2. It strengthens performance by adjusting filter properties, upgrading activation functions, and modifying convolution layers, making it more effective in identifying fake audio.
Inventors
Dr Aaeshah ALHAKAMY
inventor 3753453298_3644
Dr Jasim ALNAHAS
inventor 3753453298_3643
Dr Lubna ALHARBI
inventor 3753453298_3642
Information in this online catalogue is based on data received for the Event and updated by the participant. Catalogue entries remain the sole responsability of the participant. The Exhibition management declines all responsibility for any possible errors or omissions.