In their professional activities, investigators often have to refer to speech as a source of information. Speech is able to characterize not only the identity of the suspect or a witness, but the features of his/her mental state. However, the most important purpose of using speech analysis for the investigator is identifying false testimony. Forensic researchers found that the voice and speech of people usually differ from each other. In addition, voice and speech convey information about the territorial and social affiliation of the speaker, his/her emotional state, his/her attitude to the interlocutor, utterance and the situation as such, his/her physiological, psychological and intellectual features. Signs of speech are individual, relatively stable and repeatable. They almost do not change during the life and possess a number of specific, peculiar features. Depending on various conditions speech acquires peculiar features. Participating in the formation of traces and reflection of events, the recorded spoken word becomes a carrier of forensically relevant information needed to identify and solve crimes. By itself, it is a complex system consisting of several subsystems. In the most general sense, it can be divided into two parts: verbal and nonverbal language (facial expressions, gestures). The article discusses speech as a total of verbal characteristics. Oral speech is quite an interesting, but also complex object for study, so both the researcher and the investigator will inevitably face some problems during investigative actions. The article highlights three blocks of forensically significant problems in the study of speech: 1) the specificity of speech as a criminalistic research object which comprises: multidimensionality of speech research, modality, complex structure of speech, problem of perceiving forensically important information; 2) the situation which includes dependence of human speech on the situation of interrogation and investigator's limited powers during interrogation; 3) the subjective perception of speech as a source of forensically important information, including the problem of the ambiguity of interpretation of the investigator's questions as well as of the testimony of the interrogated and the problem of choice of the typological approach to the study of speech. The optimal typology is proposed to be the typology of accented personalities.