Generating representations of group interactions

摘要

A transcript of a group interaction is generated from audio source data representing the group interaction. The transcript includes a sequence of lines of text, each line corresponding to an audible utterance in the audio source data. A conversation path is generated from the transcript by labeling each transcript line with an identifier identifying the speaker of the corresponding utterance in the audio source data. A representation of the group interaction is generated by associating the conversation path with a set of voice profiles, each voice profile corresponding to an identified speaker in the conversation path.