[Audacity-nyquist] New vocal remover plug in developed

[David H. <dav...@ie...>]

Below is a new vocal remover that I developed as a Nyquist plug in for =
Audacity. I have also submitted it for inclusion on the Nyquist plug in =
web page.

Enjoy and let me know of any comments or recommendations, it is my first =
Nyquist plug in! Hopefully the comments in the code explain it all.


;nyquist plug-in
;version 1
;type process
;name "Vocal Remover..."
;action "Removing vocal..."
;info "Vocal Remover by David Hostetler"

;control sw "Remove in voice band or full bandwidth" int "0=3Dvoice =
1=3Dfull" 0 0 1
;control lower "Voice band lower limit..." int "Hz" 200 20 20000
;control upper "Voice band upper limit..." int "Hz" 3000 20 20000

; Vocal Remover by David Hostetler February 23, 2005
=20
; This effect is intended to remove a vocal that appears equally on the =
right=20
; and left channel of a stereo recording. The results would be used as a =
backing=20
; track for a vocalist singing the song in place of the original =
vocalist. The=20
; best cancellation of the vocal will probably be if the sound file used =
is=20
; ripped directly from the original CD, without any conversion to analog =
and=20
; back or any compression (such as MP3, etc.), either of which may =
disturb the=20
; vocal level between the right and left channel.

; Many previous versions of this function in both the analog and digital =
domain=20
; have used a simpler full bandwidth subtraction. A switch is provided =
to try it=20
; both ways for comparison.

(cond
((=3D sw 0) ; remove in just voice band
(vector
(sum (highpass8 (lowpass8 (sum (aref s 0) (mult -1 (aref s 1))) upper) =
lower)=20
(sum (highpass8 (aref s 0) upper) (lowpass8 (aref s 0) lower)))
(sum (highpass8 (lowpass8 (sum (aref s 1) (mult -1 (aref s 0))) upper) =
lower)=20
(sum (highpass8 (aref s 1) upper) (lowpass8 (aref s 1) lower)))))

; The above section subtracts one channel from the other and THEN =
filters to the=20
; voice band specified by the controls. The subtraction is done BEFORE =
any other=20
; processing because I found that if you filter before, the amplitude is =

; disturbed just a little and the cancellation is not as complete. After =
the=20
; voice band is filtered, the frequencies below and above the voice band =
are=20
; added back.

((=3D sw 1) ; remove in full bandwidth
(sum (aref s 0)(mult -1 (aref s 1)))))
=20
; This is just one channel subtracted from the other.