The embodiment of the invention discloses a subtitle generation method and device based on artificial intelligence. The method at least relates to a voice processing technology and a natural languageprocessing technology in artificial intelligence, aiming at a plurality of voice segments which come from the same audio stream and are segmented according to mute segments, texts respectively corresponding to the plurality of voice segments are obtained through voice recognition, and the time length of the mute segments is determined. And when the subtitles are determined according to the text corresponding to the target voice segment, whether the time length of the mute segment is greater than a preset time length is sequentially determined according to the sequence of the audio stream timeaxis so as to determine a to-be-processed text group comprising the text corresponding to the target voice segment. Then, a caption text is determined according to the number of characters in the textgroup to be processed and whether a separator exists or not; the text part between the separators belongs to the complete sentence and can reflect reasonable semantics, so that the possibility that incomplete sentences appear in the determined subtitle text is low, and when the subtitle text is used as subtitles to be displayed, a user watching audios and videos can be helped to understand the content of the audios and videos.