Le 14 Janvier 2006 14:51, B. Bogart a écrit :
Hey all,
How can I process a sample of a normal human voice to sound like a quiet whisper that is muffled. The words should no longer be intellegible, but it should be recognized and a voice and sound as analog as possible.
A muffled whisper? Are you sure it would then be intellegible? To muffle sound, try with a lowpass filter. To "whisperize" is not that easy.
Exerpt from http://www.csis.ul.ie/dafx01/proceedings/papers/verfaille_a.pdf "The regular whisperization is obtained by calculating the FFT of small input buffers (32 or 64 samples), and giving random values to the phase or to the modulus (cf. fig11)."
Vincent Verfaille wrote the article and have a web site about his thesis: http://adafx.free.fr/ -- Marc