Detecting emotions in voice is interesting: this content could be added to the markup of a speech-to-text conversion (here ).