Methods and systems for processing audio information to identify persons of interest in regard to a public-safety incident. An example method includes receiving human-generated narrative statements as audio and converting the narrative statements into corresponding digital texts. The method further includes determining incriminating scores for each of the narrative statements by identifying, in the digital texts, incident information classified as not being publicly known. The method further includes determining a cumulative score for an individual based on the incriminating scores tagged to the individual, making a determination of whether the individual is a person of interest based on a comparison of the cumulative score with a fixed threshold, and performing at least one automated responsive action based on the determination.