TECHNICAL DESCRIPTION
The Intellivoice voice synthesis module
produces audio speech signals when used
in conjunction with the Intellivision Master
Component (M/C) and aVoice Synthesis
game cartridge. Standard Intellivision (non-
voice) game cartridges may be operated with
the Intellivoice module but no voices will be
heard. The volume control adjusts the voice
loudness level. This control does not affect
game sound effects —only voice. The volume
level is increased by rotating the knob
counterclockwise.
Referring to the Intellivoice schematic
(Figure 5), the Intellivoice module is placed
between agame cartridge and the M/C.
Data/Address Bus Lines DB0-DB15 (carrying
game information) and Bus Control Lines
BC1, BC2 and BDIR (Bus Direction) go
directly from the game cartridge to the M/C.
This is how the M/C is able to utilize both
Voice and non-Voice cartridges through the
Intellivoice module.
In observing human speech on an oscillo-
scope, it is seen as acomplex combination
of sine waves. Digital circuits, however, know
only digital ones and zeros, ahigh or alow
voltage. Digital circuitry cannot store speech
in its analog form. Speech synthesis circuitry,
then, must be able to generate the complex
speech waveforms from adigital code. If a
spoken word were divided into, say, ahundred
parts then we could assign each part adigital
code which would correspond to its pitch,
volume, and other variables. Digital circuitry,
in this case, could generate speech from this
digitally-encoded signal.
The digitally-encoded speech segments can
come from the M/C or from memory within
the Speech Synthesizer IC (U3) itself. Speech
segments stored in U3 are in permanent Read
Only Memory (ROM), each segment address-
able when the proper address code is pre-
sented to the chip. Custom speech segments
from the M/C change with each different
game cartridge used,
Buffer/Interface IC U2 will allow addresses
of speech segments to pass onto D0-D7 (called
the Peripheral Data Bus) when Control Lines
BC1, BC2, and BDIR are in the correct state.
When custom speech segments are presented
from the M/C, U2 first converts them from
parallel (multi-wire bus) to serial (single-wire
bus). The serial data output is then loaded into
U3, the Speech Synthesis IC. This part is
controlled by control lines C1, C2, and C3,
originating in U3.
Regardless of whether the digitally-encoded
speech segments come from U2 (and the
M/C) or from the ROM within U3, the speech
segments are presented to the Vocal Tract
Model. The Vocal Tract Model consists of
electronic circuitry designed to imitate human
speech patterns, utilizing 17 different param-
eters to generate speech.
The output of Speech Synthesizer U3 is not
conventional audio, but is Pulse Width Modu-
lated (PWM). When viewed on an oscilloscope,
this appears to be a square wave whose edges
rapidly expand and contract as speech genera-
tion takes place. The signal is then sent to a
series of filters. First in Tine is C3, which acts
as a high-pass filter. This helps to limit low
frequency response to 150Hz.
Next in line are two active filters U4C/D and
U4A/B, and their associated components.
These two active filters attenuate frequencies
above 5KHz. Capacitor C8 is another low-pass
filter, and is responsible for removing most
of the "hash" sound from the PWM. The Volume
Control is next, which feeds U5A, abuffer
amplifier. Capacitor C14, connecting the
output of U5A to its inverting input, composes
alow-pass filter, further limiting the response
above 5kHz. C9 and R25 comprise the final
high-pass filter, allowing signals above 150Hz
to pass.
U5B and its circuitry is used to generate the
+1.8VDC bias used as an artificial ground
reference by the active filter networks. R7
and R8 are connected as a standard resistor
divider network so that 5V input yields 1.8 V
output at TP17. The current drain through
the divider is only 10mA, so even asmall current
drain at TP17 will upset the voltage at this
point. Since an OP Amp has avery high input
impedance, the non-inverting (+) input will
not affect the voltage at TP17.
The voltage at the +input of U5B (1 .8VDC from
the resistor divider) will force the output volt-
age to +1.8VDC. The output voltage is then
fed back to the inverting (-) input. In an OP
Amp, if the inverting input voltage changes
(such as more positive) then the output will
change in the opposite direction (less positive).
Also, the output voltage will change, trying to
force the -input to equal the +input. Say, for
example, the output voltage of U5B decreases.
Acorresponding decrease at the -input will
force the output voltage to increase until the
voltage at the -input equals the voltage at
the +input.
-3-
