Modeling a singing voice

Sound parameters:

• Singing voice, 44.1kHz (downsampled to 22.05kHz).

• block size $N=300$ $\Rightarrow$ 13.6ms/block, block overlap 50%.

• maximum of 7 partials per block.

• time stretching without significant artifacts.

• room acoustics not include $\Rightarrow$ source filter model

Use the following links to play demo sound files:

Original sound, 22kHz,     Partial model, 22kHz,     Model stretched by 2, 22kHz.