Speech to singing synthesis: Incorporating patah lagu in the fundamental frequency control model for Malay asli song

• Main Authors: Za’ba, Nurmaisara, Jamil, Nursuriati
• Format: Conference or Workshop Item
• Language: English
• Published: 2017
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://repo.uum.edu.my/22839/1/ICOCI%202017%20268-275.pdf; http://repo.uum.edu.my/22839/; http://icoci.cms.net.my/PROCEEDINGS/2017/Pdf_Version_Chap05e/PID20-268-275e.pdf

Singing traditional Malay asli music requires a type of ornamentation called patah lagu although it is often not indicated in the musical score. Therefore, new singers often learn through listening to previous performance and depend on rote learning and memory.In this paper, we introduced a new patah lagu contour to be incorporated in a speech to singing synthesis model of Malay traditional asli music.It works by modifying duration and fundamental frequency (F0) contour of input Malay spoken lyrics and convert it into singing voice based on musical notes, length and values from the score.A new contour called patah lagu based on original singing sample is incorporated in the F0 control model to accommodate Malay asli singing synthesis voice. Based on multivariate repeated measures, patah lagu makes the singing voice in the song sounds more natural as compared to without.