Synthesis of Choir Songs Using MBROLA with Multiple Voices

Previous research has explored the synthesis of singing using the input of numbered musical notation and
song lyrics, with a primarily focus on solo singers. This study takes a different approach by synthesizing songs that are simultaneously sung by 2 – 4 singers, essentially forming a choir performance. Each synthetic song a soprano (S), alto (A), tenor (T), and a bass (B), collectively known as SATB voices. Each voice type is separately synthesized using the MBROLA speech synthesis software. The resulting individual voice components are then combined into a single audio sound using the crossplatform audio editing software SOX. Findings reveal that the fundamental frequency produced when combining these four voice types differs from the average, thereby highlighting a distinctive outcome.