Preliminary study on synthesizing relaxing voices -- from a perspective of recognized/evoked emotions and acoustic features