Sonic Charge Bitspeek RE - Manual (Page 8)

Bitspeek Versus Vocoders
Vocoder is short for "voice encoder" and technically that means any device that is
capable of breaking down a voice signal into a set of parameters from which you
can reconstruct the voice intelligibly. With this terminology,
Bitspeek may be called
an "LPC vocoder". (LPC is short for "linear prediction coding" .)
However, in musical contexts, the vocoder we all know is a device consisting of a
filter bank that is controlled from a modulation source
(e.g. speech) . This filter bank
is fed with an arbitrary carrier signal (typically from a synthesizer) that is played in-
dependently of the modulation source.
This is very different from how Bitspeek works. The formant filter in Bitspeek is not
implemented as a bank of bandpass filters. Instead, the analysis is divided into time
frames
(typically around 10ms each) . For each frame the algorithm quickly con-
structs a resonant filter that closely represents the formants of the input speech.
In other words, whereas conventional vocoders works with a number of bandpass
filters at stationary frequencies,
Bitspeek creates animated filters whose peaks are
precisely positioned in the spectrum. On the other hand, the signal is chopped up in
discrete time frames, whereas the conventional vocoder varies the formants
smoothly over time.
Example Combinators
Bitspeek for Rack Extensions ships with some example Combinator patches. They
were created to be inspirational starting points for further exploration of the inter-
connectivity possibilities in Reason. Here is the run down with screenshots and ex-
planations.
ToyTalk (plain)
This is a plain single Bitspeek only, mainly provided as a reference point. It is config-
ured to produce a sound reminiscent of a certain early 80's toy.
SubSpeek
One Bitspeek controls the pitch
and the volume of a SubTractor,
which in turn is routed back into
Bitspeek for filtering. In effect, the
SubTractor will replace the built-in
synth in
Bitspeek , enabling com-
pletely new timbres.
Notice how we have connected
both
NOTE and PITCH CV OUT to
achieve exact pitch tracking.
VOICE VOLUME OUT to "Amp
Level" will make the SubTractor
play voiced sounds only
(and not
noise) .
8! ! © 2012 Sonic Charge