Robotic Mouth Replicates Human Singing

Welcome the latest singing sensation from Japan! A robotic mouth was demonstrated at Robotech 2011 that not only talks but can also sing. Hideyuki Sawada, Professor from Kagawa University displayed his creation by making the “mouth” to recite a Japanese children’s song, “Kagome, Kagome.” The mouth was able to replicate the singing with the help of vocal cords, vocal tract nasal cavity, auditory system and lung trachea.

#-Link-Snipped-#
Robots can sing too!

The vocal tract consists of a resonance tube, tongue, vocal chords and even a nasal cavity. The big horn like structure above lips that you see in the picture is the nasal cavity. The mouth learns while it sings. A microphone placed just in front of the lips analyses the sound it creates and gives these signals to the learning control block. The sound is fine-tuned continuously by a neural network working in background, getting the signal of target sound from sound analyzer and feeding it to the vocal tract model. This is termed as the running phase of the singing. In the learning phase of the singing, the sound signals are controlled and sent to the neural network depending upon the parameters in the past. In short, the mouth learns from its mistakes and tries to rectify them giving better sound quality each time.



#-Link-Snipped-#
Detailed intricacies of the singing mouth.

The mouth is replete with around 8 resonance control motors that move the vocal tract and the lips to sing. Same is the case with humans. We control our muscles while forcing air with the help of lungs. The different notes are created by the change in frequency of the sound wave. This change is brought about by the vocal chords. Similarly in this robotic mouth, the air is forced through the artificial vocal chords by an air compressor. There is an intermediate pressure valve to control the flow of the air from compressor to the vocal chords. The pressure of air controls the pitch of the sound, for which a separate pitch control motor is provided.The tongue is made of silicon and can move inside the vocal tract.


The developers have tried to make the mouth exactly similar to that of humans in technical aspects. The robotic singing sensation is not new. There have been attempts to recreate the human singing and talking in the past, such as the Voxhead robot. However, the mouth developed by Professor Hideyuki is more intricate and detailed as far as technical aspects are concerned. The attempt to replicate the human mouth is better (still, teeth are absent!) in this case. Although the developers have tried their level best to come up with a model that is technically sound, it seems they really need a plastic surgeon it more attractive. For now, Justin Bieber does not seem to have any threat from Professor Hideyuki’s pupil. Till then lend your ear to “Kagome Kagome."

Source: #-Link-Snipped-# and #-Link-Snipped-#.