Supervisor
Prof. Sten Ternström, TMH Division of Speech, Music and Hearing, KTH
Dates
Nov. 2021 - Jan. 2022
Embedded Files
Singing synthesis with spectral notches
Singing synthesis with spectral notches
@KTH Royal Institute of Technology, Course DT2217 Advanced Individual Project in Music Acoustics
A source-filter based singing voice synthesizer with an intuitive GUI, implemented with SuperCollider. The synthesis model incorporated notch filters to produce anti-resonances in addition to formants, which improved the perceptual quality of the voice synthesis.
This implementation was later incorporated into the new KTH Music and Singing Synthesis Equipment (MUSSE) system.
My role
Audio programming
UI development
Perceptual Study
Technical report
Spectrum of vowel /a/. Blue: real voice, green: synthesis with formants, red: synthesis with formants and notches.
Spectrum of vowel /e/. Blue: real voice, green: synthesis with formants, red: synthesis with formants and notches.
Spectrum of vowel /o/. Blue: real voice, green: synthesis with formants, red: synthesis with formants and notches.
Page updated
Google Sites
Report abuse