Please use this identifier to cite or link to this item:
|Title:||Singing voice synthesis||Authors:||Tan, Hwee Teng.||Keywords:||DRNTU::Engineering::Electrical and electronic engineering||Issue Date:||2011||Abstract:||Singing voice synthesis has resulted in beneficial commercial usage such as the VOCALOID Singing Synthesis Software by Yamaha. However, in order to reduce the complexity in the synthesis process and hence reduce cost, a closer look into the relationship between pitch and formants is necessary. From this research, a conclusion that formants are changing with pitch and most suitably described as linearly changing can be drawn. With the above conclusion on the relationship between pitch and formants, further testing was conducted through hands-on programming using MATLAB to conduct gender transformations on recordings. Recordings generated were then analysed and compared to determine the successfulness and efficiency of a total of four different spectral shifting functions. Gender transformation was the only tested transformation and results generated tallied with previously found relationship between pitch and formants. The linear function produced the best result out of the four functions. As only one transformation was tested, in the near future, more transformations can be tested with more variety of spectral shifting functions.||URI:||http://hdl.handle.net/10356/44656||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.