The interaction of participants in digital platforms requires a prompt response to actions from each other. At the same time, such interaction is multifaceted, in which, along with the transfer of information, knowledge about the state of the interlocutor, his current emotional state and personal characteristics is necessary. As part of a pilot study at ITMO University, together with psychologists from St. Petersburg State University, a list of markers of aggressive behavior was determined. And within the framework of this work, the implementation of methods for extracting markers of aggressive human behavior through audio and text channels is described. This paper also describes the approach of combining verbal and non-verbal features in speech. From the results obtained in the course of applying methods for extracting markers of aggressive behavior, we can conclude that the use of audio and text channels of information separately is not enough to detect aggressive behavior.