日本造美女机器人 竟能面带表情唱歌

　　Lady Gaga will not be losing any sleep. But this eerie humanoid robot has been taught to sing just like a real pop star. The life-sized robot, known as HRP-4 , has a realistic face and movable features that are able to mimic the expressions of a human singer. But she is now able to 'sing' using a synthesized voice technology that sounds and breathes like a human too.

　　唉，这下嘎嘎小姐可以稍微休息一会了，原来这位来自日本型号为HRP-4的“人型机器人”竟然能代替人类嗓音唱出“天籁般”歌声。据悉，新开发的机器人HRP-4拥有着与人类近似的面孔，就连模仿人类唱歌都能演绎得“惟妙惟肖”。科学家应用高科技仿真手段将这位“天使般面孔”的机器人唱歌时“换气技术”做得与人类一模一样。

　　Masataka Goto, from Japan's Institute of Advanced Industrial Science and Technology, said: 'This robot primarily utilises two technologies. One is for its singing voice. 'Last year, we used Yamaha's Vocaloid software to synthesize notes directly. But this year, we've used our own technology, called VocalListener.This is a new technology, which synthesizes a singing voice on the computer side, as if imitating a person's singing voice.

　　来自日本科学技术学院的专家说，型号为HRP-4仿真日本动漫机器人的最大优点就是拥有着和人类“极为相似”的歌声，这是通过应用新型电脑技术 VocalListener来实现的，发音与人类几乎完全相似。

　　'Also, for the robot's facial expressions, we've developed another new technology, called Vocawatcher. This analyses a video of a person singing, to create expressions naturally.' Researchers used a real singer as a model, recording her every move as she sang a typical Japanese pop song. The movements were mapped onto HRP - 4 who was then able to mimic her real-life counterpart's every movement.The team even helped the robot breathe realistically, by modeling real, human breathing sounds.

　　此外，这款仿真日本动漫机器人还在“面部表情”上加以改进，应用了Vocawatcher技术将人类唱歌时候的自然的面部表情完全“拷贝”在机器人的脸上。研究人员称，他们将真人模特做实验，让真正的歌手唱歌，同时记录歌手在唱歌时的“脸部细节”，然后，所有的脸部细节都会自动传导到HRP-4仿真日本动漫机器人的系统中，让机器人也拥有和真正歌手一般的“自然唱腔”。