tebop opened this issue on Jun 10, 2003 ยท 7 posts
ockham posted Tue, 10 June 2003 at 12:36 PM
Singing is essentially different from speech. The vocal folds and mouth are shaped in special ways for singing, which produces a distinct set of frequency components in the resulting sound. A spectrum analyzer like Mimic is designed to separate the formants (frequency components) in normal speech, so it will naturally make mistakes when trying to deal with singing. And if there are instruments in the background, the task becomes just about impossible.