Spoofing Detection of Fake Speech Using Deep Neural Network Algorithm
Kristiawan Nugroho, Edy Winarno · 2022
Spoofing is a challenging research topic in Speaker Recognition. Spoofing, among others, can use fake speech, especially in the form of voice identity falsification or fraud, which is a problem that must be resolved. Various classification methods in Data Mining have been used in research to detect spoofing. However, The low level of accuracy, especially in managing large data, is an obstacle to using this approach. Deep Neural Network (DNN) is one of the methods in Deep Learning that is often used in research that processes extensive data. The DNN approach is proven to have good performance. This study uses the DNN method in detecting the authenticity of the speaker's voice. The results show that DNN is a method that has good performance in detecting fake speech spoofing with a model accuracy rate of 96.5%, 97.3% precision, 96.5% recall, and 96.7% F1 Measure.