Annotation: This paper describes a study of the problem of generalizing multimodal data in the detection of artificially synthesized audio files. As a solution to the stated problem, a method is proposed which combines simultaneous analysis of audio file characteristics with its semantic component presented in the form of text. The approach is based on graph neural networks and algorithmic approaches involving the analysis of keywords and text sentiment. The conducted experimental studies confirmed the validity and efficiency of the proposed approach

Keywords: deepfake, graph neural networks, artificially synthesized audio file, text analysis