SpecDraw
SpecDraw is a prototype implementation of a sonagram-based frequency-domain
sound editor. The program was developed by Gerhard
Eckel at IRCAM in 1989 on a NeXT
computer system and was first presented at the 1990 International Computer
Music Conference in Glasgow (Eckel
90). The main purpose of SpecDraw was to proof the significance of
a new spectral editing paradigm based on interactive design of time-variant
filters by drawing directly on a frequency-domain representation (Eckel
92). This intuitive signal editing metaphor was very well received
by composers since it allows for sophisticated signal editing without specialized
knowledge in signal processing. The combination of visual and auditive
control over signal editing tasks also enhances the understanding of acoustic
phenomena and their perceptual relevance, which is a prerequisite for an
artistically meaningful employment of digital signal processing in computer
music. The experience made with SpecDraw served as basis for the development
of AudioSculpt,
the graphical user interface for SuperVP
running on Macintosh computers.
Example
The
example shows how a singer's voice can be extracted easily from a recording
which contains both the singer and the orchestra. In the sonagram representation
(to the right) the partials of the singer can be easily distinguished from
the orchestra instruments due to the pronounced vibrato of the voice. In
order to separate the voice from the orchestra, all the user has to do
is to select the relevant partials by drawing regions on top of the sonagram.
SpecDraw uses this information to construct a time-varying filter which
it applies to the original signal using an efficient FFT-based filter algorithm.
The result is a signal which contains only the partials of the voice.
The
example shows how a singer's voice can be extracted easily from a recording
which contains both the singer and the orchestra. In the sonagram representation
(to the right) the partials of the singer can be easily distinguished from
the orchestra instruments due to the pronounced vibrato of the voice. In
order to separate the voice from the orchestra, all the user has to do
is to select the relevant partials by drawing regions on top of the sonagram.
SpecDraw uses this information to construct a time-varying filter which
it applies to the original signal using an efficient FFT-based filter algorithm.
The result is a signal which contains only the partials of the voice.