The research aims to identify the most significant differences in the speech characteristics of consecutive translations performed using speech recognition tools and traditional note-taking. The paper conducts an analysis of a corpus of empirical data, including the script of an English-language speech and the scripts of consecutive translations of this speech into Russian. While there is anecdotal evidence of using speech recognition results as a crutch for consecutive interpretation, there is a lack of scientific research devoted to this topic. The novelty of the research lies in its first-time description of the impact of speech recognition tools on the characteristics of consecutive interpreting. It has been found that relying on speech recognition results statistically significantly increases the average sentence length in translations and affects the selectiveness of the interpreters’ attention, which gradually increases as the interpreting process progresses. The research shows that translations performed using speech recognition results are closer to the linear structure of the original message in syntactic and semantic terms. As a result, such translations contain more semantic details, which are not equivalent in terms of completeness and accuracy of translation. The results of the analysis suggest that interpreting with the support of recognized speech reduces the cognitive load during the production of the target message.