Disclosed embodiments provide techniques for suggesting images that can be conveniently included in electronic communication such as social media posts, emails, text messages, and other forms of electronic communication, based on voice input and/or emotion analysis from facial expressions of a user. A speech-to-text process converts a user's spoken utterance to text. Natural language processing performs an entity extraction on the utterance to generate a scenario summary. An image database is queried using the scenario summary and/or emotion analysis based on user facial images. A mapping rule library may be used to perform the image database query. Based on the scenario summary and/or the emotion analysis, images are presented to the user as recommended for including in an electronic communication message such as a social media post, email, or text message.