Micromint MICROVOX User manual

USERS MANUAL
THE MICROMINT INC. 561 Willow Avenue, Cedarhurst, N.Y. 11516

TABLE OF CONTENTS
Page
Introduction 1-2
MicroVox Hardware 3-4
Speech Synthesizer and Inflection Circuitry 5-6
Device Control Codes 7-15
Setting the Serial Port 15-16
Parallel Input Port - 17
Phoneme Codes 18-19
J3 - Connector Pinout 20
Jl - Connector Pinout 21
Phonetic Word List 22-25
Parts List 26-27
Layout 28
Schematics 29-32
Warranty 33

THE MICROVOX TEXT TO SPEECH SYNTHESIZER
Copyright 1982 The Micromint Inc.
The MicroVox is a stand alone intelligent peripheral that
converts ASCII text to spoken English. The MicroVox is attached
to the computer (or terminal, modem, etc.) via either a serial
or parallel cable. Its operation is similar to that of a
printer except that the output is speech rather than printed
word. The MicroVox has many programmable options which produce
its high level of intelligibility. It has the following
features:
* Phoneme based speech synthesizer
* 6502 Microprocessor
* 64 crystal controlled inflection levels
* 700 character buffer (optional 2.7K)
* 6K byte text-to-phoneme algorithm
* Full ASCII character set recognition and echo
* Adjustable baud rates (75-9600)
* EIA RS232C and parallel input interfaces
* Phoneme access modes
* Serial X-on/X-off handshaking
* One watt amplifier and volume control
* Onboard power supply
* Music and sound effects capability
Basically, the MicroVox Speech Synthesizer consists of a
6502 based microcomputer with a voice synthesizer output port.
It has a 6502 microprocesor, crystal controlled 75-9600 bps
full handshaking serial interface, parallel input port, 2K
bytes of RAM, 8K bytes of EPROM, and has an onboard power
supply. The EPROM contains the operating system and text to
speech algorithm.
Special control signals are sent from the host computer to
select among many different user options. In general, these
control signals are in the form: I(letter)(option). The
exclamation point is a signal to the MicroVox that a control
code follows. Options can be changed at any time by sending the
appropriate codes preceding or imbedded within the text.
What is a Text to Speech Synthesizer ?
With the majority of speech synthesizer interfaces, text to
speech, or the actual conversion from ASCII characters to
phonemes, LPC formants, word codes etc., is left to the user.
Such a conversion routine will be more or less elaborate,
depending upon the required vocabulary. For short vocabularies,
the conversion program might consist merely of a table of words
and their appropriate synthesizer codes. When the required
vocabulary becomes very long, or in fact unlimited, tables
become cumbersome and a text to speech algorithm is required
instead.
page
1

A text to speech algorithm is a program which takes ASCII
data and performs a synthesis by rule analysis of character
strings. It determines which characters are voiced and which
are silent by following a set of general rules for pronouncing
English (text to speech algorithms can be written for other
languages as well). Text to speech algorithms vary in length
depending upon exactness of voice translation. Typical
algorithms are in the 4K to 8K byte range but f some of the more
sophisticated programs are up to 80K bytes. The primary
difference between a 6K and a 20K algorithm is more often the
spelling of input text rather than any specific sound quality
differences (an 80K algorithm can often be half look up tables
for exceptions to the rules). For exact pronunciation it might
be necessary to spell words differently to more easily fit the
prescribed rules on the smaller algorithm, such as entering
"com pu ter" instead of computer. The only other limitations
are features such as pronunciation of punctuation or inflected
speech. Both of these capabilities are supported in the
MicroVox.
The MicroVox text to speech synthesizer is a smart
peripheral. It speaks only those ASCII strings which are
directed to it through either its serial or parallel input
ports. The ASCII text can result from PRINT statements in BASIC
or the contents of complete disk files. MicroVox connects to
the computer in the same manner as a printer or modem and
virtually anything that can be printed or viewed on the CRT can
be vocalized.
The MicroVox is a combination of two major elements: a 6502
based microcomputer and a Votrax SC-01 speech synthesizer chip.
The SC-01 is a CMOS (complimentary metal oxide semiconductor)
chip which consists of a digital code translator and an
electronic model of the vocal tract. Internally, there is a
phoneme controller which translates a 6 bit phoneme and 2 bit
pitch code into a matrix of spectral parameters which adjusts
the vocal tract model to synthesize speech.
The output pitch of the phonemes is controlled by the
frequency of the clock signal. The clock frequency is nominally
720 KHz but subtle variations of pitch can be induced to add
inflection. This prevents the synthesized voice from sounding
too monotonous or "robotlike".
Listed in Table 1 are the 64 phonemes defined for the
English language (three produce no sounds). The phoneme sound
is generated when a 6 bit phoneme code is transmitted to the
SC-01. Each phoneme is internally timed and has a duration of
47-250 msec depending on the particular phoneme selected and
the clock frequency. The computer operating system sends these
codes to the synthesizer chip through a latched parallel output
port and monitors the synthesizer's activities (the A/R line)
through an interrupt line.
page
2

The MicroVox Hardware
As previously mentioned, the MicroVox is a stand-alone
microcomputer configured to function as an intelligent
peripheral. Figure 1 is a basic block diagram of MicroVox. It
can be viewed as a general purpose 6502 based computer with a
speech synthesizer attached as a memory mapped I/O port.
MicroVox is best explained by dividing the circuitry into
four functional subsections: processor and timing, memory,
serial and parallel I/O, and speech synthesizer. Figure 2 is
the complete MicroVox schematic.
Processor and Data Rate Clock
The processor is a 1 MHz 6502. The processor and data rate
clocks are derived by dividing down a 4.9152 MHz crystal
through IC6. Using a 4.9152 MHz crystal (base is 75 times 2 to
the 16th) and a 12 stage CD4040 binary divider (IC6), 9 rates
are derived directly: 75 bps, 150 bps, 300 bps, 600 bps, 1200
bps, 2400 bps, 4800 bps, 9600 bps, and 19200 bps (while the
hardware can produce 19200 bps, it is not supported by the
operating system). The MicroVox will not communicate at 110
bps. See "Setting the serial port". The 6502 processor operates
at a clock frequency of 611 KHz.
Memory Section
ICs 2-5 and 9 form the address decoding and memory section
of the MicroVox. IC 9 decodes the 5 most significant address
bits to create 8 strobes. They are defined as follows:
Name
Hex Address
Connection and Function
SEL0 000 IC2 Memory Block (RAM)
SELl
800
IC3 Memory Block (RAM)
SEL2
1000 IC10 Serial Port
SEL3 1800
IC11 Parallel Ports
SEL4
8000
IC14 Inflection Clock Rate
SEL5
A0 00
IC14 Phoneme Latch
SEL6 cooo
IC5 Memory Block (EPROM)
SEL7 E000
IC4 Memory Block (EPROM)
the
MicroVox configuration, ICs 2 and 3 are intended
be RAM while ICs 4 and 5 are EPROM or ROM. The pin designations
for ICs 2 and 3 are for 2K by 8 RAM chips such as the Hitachi
6116 or Toshiba 2016 (these devices are 2716 pin compatible.
You could also use 2716 EPROMs in these sockets). This
programmable memory is used for conversion tables, register
stacks, and the ASCII input buffer (the MicroVox can receive
data faster than it can speak it). The basic MicroVox uses only
one RAM chip which allows a 700 character input buffer. By
adding the second RAM chip in IC3 (and changing a few EPROM
constants), this buffer can be optionally expanded to 2.7K
characters.
page
3

The text to speech algorithm is placed in EPROM/ROM
positions ICs 4 and 5. Either 2716 (2K X 8), 2732 (4K X 8), or
2764 (8K X 8) devices can be used in these positions depending
upon the jumper selections JP4 and JP5. The 8K byte MicroVox
software will be either on two 2732 EPROMs and require both
sockets or a single 8K 2764 (or ROM equivalent).
Serial and Parallel I/O
MicroVox, unlike most other voice synthesizers, has both
serial and parallel input ports to receive ASCII characters.
The serial port uses a 6850 asynchronous communications
interface adapter (ACIA, IC10) which is software programmable.
During initialization, the ACIAs functional configuration is
preset. Considerations such as word length, clock division
ratios, parity, stop bits, etc., are selected by properly
setting bits in the ACIA's control register. The data rate is
set by the system data rate clock (from SW2 and IC6) and data
is sent and received from the Transmit and Receive data
registers respectively. Information such as framing errors,
parity errors, and buffer and handshaking status, are
determined by reading the ACIA status register.
On the MicroVox, the serial port can be used with or without
hardware handshaking (CTS, DCD, RTS, etc.). This is especially
useful when communicating over modems or terminals which have
no handshaking signals. Instead, the MicroVox software
incorporates software handshaking.
When receiving ASCII text in the software handshaking mode
the MicroVox sends an to the host computer when its input
buffer is almost full (the host should stop sending data). It
sends a "#" when it is ready to receive data again. Obviously,
even this can be ignored if the data rate from the host
computer never exceeds the speed at which the buffer is
emptied.
The parallel input section uses an 8255 PIA (IC11) which is
also programmable. As configured, 8 bits of it are used to
receive parallel format ASCII data such as would be transmitted
to a parallel printer. Using 2 additional pairs for the strobe
and acknowlege handshaking, the MicroVox can attach to any
conventional Centronics printer interface. (As configured, the
34 pin edge connector is exactly compatible with the Radio
Shack TRS line of computers and can connect directly to their
34 pin Centronics printer edge connector).
DIP switch SWl also attaches to IC11. Switch positions 6
thru 8 set serial word length, stop bits, and parity on the
ACIA; switch section 3 selects hardware or software
handshaking; sections 1, 2, 4, and 5 are not used.
page
4

Speech Synthesizer and Inflection Circuitry
Probably the most important section of the MicroVox is the
actual speech synthesizer circuitry. The MicroVox allows 64
levels of pitch inflection.
The output pitch of the phonemes is controlled by the
frequency of the clock signal. The output pitch is a function
of this clock input frequency and two pitch control lines, il
and 12 (each acts independently). Four rather large variations
in pitch (corresponding to !P1 thru !P4 in the operating
system), can be achieved simply by using these manual
inflection inputs. More subtle variations in output pitch are
attained by externally controlling the synthesizer clock. Using
the 1.22 MHz system clock and a digital rate multiplier, a
programmable clock can be created to produce smaller and more
defined pitch inflection changes.
On a SEL4 strobe, a four bit inflection code is latched into
IC13 and applied to the rate multiplier. The four bit
combination results in 16 clock rates from 614.4 KHz to 902.4
KHz in 19.2 KHz increments (corresponding to !Rl thru 1R16 in
the operating system). 20 KHz creates a relatively small pitch
change by itself (out of a 720 KHz nominal input frequency)
but, used dynamically in a sentence it creats a definite
improvement in intelligibility.
The pitch levels IPl thru IP4 are the base pitch and the 16
frequencies from the rate multiplier, IRl thru 1R16, are the
clock rate. The combination of the two functions results in 64
pitch levels or inflections.
The pitch at which individual phonemes are pronounced may be
controlled automatically by the text to speech algorithm, kept
fixed, or altered by user command. Some users prefer automatic
inflection, because of the variety it gives to the speech.
Others think a computer should sound like a computer and prefer
the flat speech to artificially intoned speech. Still others
may wish to directly control the pitch to make the unit "sing"
(pitch and rate codes may be mixed with phoneme codes to
produce "singing") or pronounce words with special emphasis.
The user may control the base pitch setting independently of
the clock rate. The user options are:
IPl (low pitch)
IP2 (medium low pitch)
IP3 (medium high pitch)
1P4 (high pitch)
The user may also control the clock rate.
IRl (slowest rate — lowest level for the given base pitch)
IR2 (slightly faster)
1R3...IR16 (increasingly faster rates)
page
5

The MicroVox has the ability to play musical notes and
produce sound effects. This is accomplished by using a program
routine to toggle one bit of IC11 at a predetermined rate. This
lead, in addition to the output from the speech synthesizer
chip (IC12) is connected to the output amplifier. The results
are similar to the sound produced on the internal speaker in an
APPLE II computer (it uses the same technique).
OPERATOR INTERACTION WITH THE TEXT TO SPEECH SOFTWARE
The MicroVox is a stand alone intelligent peripheral that
converts ASCII text to spoken English. The MicroVox is attached
to the computer (or terminal, modem, etc.) via either a serial
or parallel cable. Its operation is similar to that of a
printer except that the output is speech rather than printed
word. The MicroVox has many programmable options which produce
its high level of intelligiblity. These options are called
device control signals and are transmitted to the MicroVox
along with the text.
Device controls signals are sent from the host computer to
select among many different user options. In general, MicroVox
control signals are in the form:
l(letter)(option)(option)
for example: 1HXY
The exclamation point is a signal to the MicroVox that a
control code follows. The user may if he wishes use any other
character as the signal. This is done by giving the following
instruction:
(old signal character)X(new signal character)
for example: 1X$
will change the control signal from an exclamation point to a
dollar sign and $X* will change it then from a dollar sign to
an asterisk.
Device control signals can be imbedded anywhere in the text
transmission and are not spoken. Once a device control signal
has been sent to the MicroVox, all succeeding text entry will
be subject to that default setting until it is changed. For
example if letter by letter pronunciation is invoked with IE
then all text will be spelled until a IT is sent to reinvoke
text to speech translation.
page
6

DEVICE CONTROL CODES
Software Handshaking
If standard parallel or RS-232C serial connections are used
the sending computer hardware will detect and examine the RTS
signal and determine whether the MicroVox is ready to receive a
character or, if busy, take appropriate action. However, many
popular brands of microcomputers lack the hardware to detect
RS-232C handshaking signals and these handshaking signals do
not pass through modems back to mainframe computers. In the
MicroVox, special software handshaking signals, described
below, are provided for these purposes (in general, hardware
handshaking is preferable whenever it is possible to use it,
because it relieves the host computer's processor of the
handshaking chores and allows use of higher data rates).
For software handshaking, switch position 3 on dip switch
SWl is set in the closed position (open is hardware
handshaking). The following option is provided:
!H(busy character)(ready character)
Example: !H@#
In the example shown, the MicroVox will send the character
to the computer when it is unable to receive more data, and
will send to the computer when it is again ready to receive
data. It is the responsibility of the computer programmer to
write the software necessary for the use of these options.
NOTE: While in the example above the handshaking characters
are '@' and '#', the default mode of the MicroVox uses the
characters 'R' and 'B' instead. Use the above described method
to set any other pair of handshaking characters.
Finally, it is possible to use the MicroVox with no
handshaking by simply invoking the software handshaking mode
and ignoring the handshaking transmissions. In this case, it is
the user's responsibility to insert timing delays in the
program so that data will not be sent to the MicroVox faster
than it can handle the data.
Speech, Spelled Speech, Phoneme Code, and Music Modes
The MicroVox can operate in four different modes: text to
speech, text to spelled speech, phoneme codes, and music. When
the MicroVox is turned on it is in text to speech mode,
however, the user can select among the following options:
IT (text to speech)
IE (spelled speech — say each letter)
1C (phoneme codes)
IN (musical notes)
Page 7

NOTE: The default mode is IT. To exit any mode you must enter
another. For example, if you are in the IE mode, to return to
text to speech you must type IT. Also, changing between mode
frequently resets selected options to the default mode.
Text to Speech
The software used in the text to speech algorithm
incorporated in the MicroVox is derived from an algorithm
conceived by the Naval Research Laboratory. This algorithm
combines word, morph and letter rules in a single table of
about 400 rules. This table contains subtables for each letter
of the alphabet and achieves very intelligible speech.
In the text to speech mode (IT), this algorithm attempts the
correct pronunciation of any phrase sent to it. However, no
program of reasonable size can possibly contain all the rules
and exceptions for the pronunciation of English. Moreover,
since the MicroVox lacks extra-sensory perception, it cannot
tell for instance, when the user sends "READ" if the present or
the past tense is meant. The solution when a word is not
pronounced to the user's satisfaction is to alter the spelling.
By typing RED or REED instead of READ, the user can be sure to
get the desired pronunciation. If HICCOUGH is pronounced
strangely, try HICCUP. Often it helps to break a word into
syllables. Compare the pronunciation of TYPEWRITER and TYPE
WRITE ER. Foreign words will require considerable ingenuity,
since the MicroVox works on the principles of English
pronunciation. Compare PARLEZ VOUS and PARLAY VOO.
Spelled Speech
The spelled speech mode is useful for abbreviations and
words that a user might have difficulty in understanding. When
this option is selected, every letter is pronounced separately.
(By selecting the IA punctuation mode, punctuation will also be
pronounced).
Example: IT THE WORD AWFUL IS SPELLED IE AWFUL IT
In this example, the MicroVox will say "THE WORD AWFUL IS
SPELLED", and then spell out A W F U L. The IT at the end
returns the Microvox to the text to speech mode.
Phoneme Mode
The MicroVox may also be programmed directly in phoneme
codes. A space must be left between the mnemonic codes.
For example:
IC AE N D PAO THV UH2 PAO S E PAO I Z PAO B 01 AY 13 L I
NG PAO H AH T PAl
will say "and the sea is boiling hot".
Page 8

The intonation I or F modes can be either on or off when
phoneme codes are used. If the intonation is off, the rate
which is output will be the base rate. If it is on, intonation
will be like that for text. If there are errors in the codes,
the erroneous codes will be spoken as if they were text.
Music Mode
Music mode can be turned on by IN. In music mode, the
following notation is used. There are 7 octaves centered about
middle C, indicated by numbers from 1 to 7. Notes are A, B, C,
D, E, F, G. A sharp is indicated by "+", flat by The
length of a note may be from 1 to 256 times an internal time
constant. Rests are indicated by R. For instance 3F+26 means
third octave, F sharp, 26 time constants long. R16 means a
sixteen time constant rest.
The music mode suspends the MicroVox operating system and no
serial or parallel data can be received during music output.
Also, entering music mode will reset most previously set
control codes.
Text Synchronization
For many applications it is important to synchronize speech
with external such as text or actions appearing on the screen.
For instance, an instructional program may require placing a
picture on the screen when certain speech output begins and a
question on the screen when it ends. For synchronization, the
following option is provided:
IK(synchronization character)
Example: IK#J0HN!K%MARSHA1K$
In the example shown, the MicroVox will send a "#" back to
the computer just before starting to say "JOHN"; it will send a
"%" to the computer just after saying "JOHN" and just before
starting to say "MARSHA"; and it will send a "$" character to
the screen just after saying "MARSHA".
Example: LOOK AT THE SCREEN NOW IK#
In this example, a "#" will be transmitted to the host
computer after saying "LOOK AT THE SCREEN NOW".
None of these special synchronization characters will be
spoken. It is the programmer's responsibility to use the
incoming synchronization characters to coordinate the screen
display with the speech.
Page 9

Phrase Termination
Many aspects of English pronunciation are controlled by the
context in which a given letter or word is spoken. For this
reason, the MicroVox will await a complete phrase before
translating from text to speech. If the user does not specify
otherwise, the MicroVox will wait to translate a phrase until
it has received one of the following phrase terminating
characters:
(1) a period followed by two spaces or a carriage return
(2) a comma, semicolon, colon, exclamation point, or
question mark followed by a space or carriage return.
(3) a carriage return
For some types of output, such as computer programs or
poems, the user will want each line read as a separate phrase.
For others, such as ordinary English text, the user may not
want a carriage return to terminate a phrase. The user is given
the following options to deal with this situation:
!L and IW
"IW" means "Whole text pronunciation". If this option is
selected, a carriage return will not terminate a phrase unless
the carriage return is preceded by one of the punctuation marks
indicated in (1) and (2) above.
"!L" means "Line-by-line pronunciation". If this option is
selected, a carriage return will always be treated by the
MicroVox as terminating a phrase. When the MicroVox is first
turned on it is in the "L" mode.
Rather than send a special signal to terminate a phrase, the
user may wish to have the MicroVox treat a phrase as terminated
if a certain delay occurs without any phrase terminator being
received. Possible applications of this option include
situations where the user does not fully control the output.
For instance, suppose the MicroVox is passively connected to a
transmitting device which doesn't send any of the terminating
characters listed above (maybe it sends "STOP" instead). In
such a case, there is no way to insert phrase termination
characters in the output stream. However, if the MicroVox is
set to treat a half second delay without receipt of information
as the end of a phrase, computer output will not be lost or
ignored.
The user is given the following option to provide delayed
phrase termination:
ID(delay number)
1D1 through ID8 result in a delay of 50 x 2n milliseconds
where "n" is the number following "D" (Note: If too short a
delay is used, a phrase may be translated in pieces resulting
in odd intonation or pronunciation, since the MicroVox uses the
context of letters and words to determine their pronunciation.)
Page 10

1D9 is a special case. The MicroVox waits for a phrase
terminating character even if it has to wait forever. 1D9 is
the default mode (at power up) and should be used with slow
data sources such as hand typing on a terminal.
This selectable delay feature is particularly useful for the
handicapped. It allows a blind programmer to use a standard
unintelligent terminal. This is facilitated by connecting the
MicroVox to receive the output from both the user and the
computer. Using the "ID" command, the MicroVox can echo all
communication either way. If the delay is set to about 0.1
seconds, keys pressed by the user would be echoed as spelled
letters because the slight delay between them will be treated
as an end of phrase but, output generated by the computer will
be spoken as complete lines, because there generally will be no
significant delay between characters. The delay may be varied
to fit the particular application.
The MicroVox must be in the IF mode before entering the D
mode. Also, once in the D mode, other control changes can only
be received if the MicroVox is set to 1D9 first (so that it can
interpret the input rather than just echo the characters).
Intonation
Within the MicroVox, a special intonation algorithm is
included. However, providing realistic intonation is much more
difficult than choosing the correct phonemes. Most intonation
patterns are not represented in English spelling. Without
knowing the writer's state of mind, achieving the proper
intonation may require grammatical parsing of a sentence. The
algorithm attempts to raise the pitch on stressed syllables,
raising it at the start of sentences and before commas,
lowering the pitch before the period at the end of a sentence.
Before a question mark, the pitch is ' raised, unless the
sentence begins with a question word (who, what, when, where,
etc.), in which case it is lowered.
The pitch at which individual phonemes are pronounced may be
controlled automatically by the text to speech algorithm, be
kept fixed, or be altered by user command. Some people prefer
automatic inflection, because of the variety it gives to the
speech, even though the inflection is often not accurate.
Others think a computer should sound like a computer and prefer
the flat speech to artificially intoned speech. Still others
may wish to experiment with controlling the pitch themselves to
optimize intelligibility. This control can extend even to make
the MicroVox "sing".
The hardware in the MicroVox allows control of pitch in two
different ways. The VOTRAX SC-01A synthesizer chip has four
selectable pitch levels. In addition, the output pitch may be
varied by selecting one of sixteen different rates for the
clock which controls the synthesizer chip. When the MicroVox is
first turned on, the synthesizer chip is set to base pitch
level 1 (low) and clock rate #5 (defined below). The intonation
Page 11

is generated by an algorithm which selects an appropriate clock
rate for each phoneme. To turn on or off the automatic
intonation algorithm, the user may send the command:
IF (flat intonation — monotone)
and the output rate will stay at the default base and clock
rate. To invoke automatic clock rate setting, the user may send
the command:
II (inflected intonation by algorithm)
The intonation algorithm adds or subtracts from the base
rate to ultimately select the final voice pitch. Using the II
mode however, only four clock rate pitch level shifts (out of
16 possible) are used.
The user may decide not to implement automatic inflection on
all text to speech translation yet desire to add certain pitch
changes on specific words or phonemes. This can be easily done
on the MicroVox since the base pitch and the clock rate can be
controlled independently and changed at any time. The user
options are:
IP1 (low pitch)
IP2 (medium low pitch)
IP3 (medium high pitch)
IP4 (high pitch)
The user may also control the clock rate:
IRl (slowest rate, lowest level for the given base pitch)
IR2 (slightly faster)
1R3...IR16 (increasingly faster rates)
Example: IPi IR5 THIS IS A IR8 TEST
In this example, "THIS IS" will be spoken at clock rate R5
and "TEST" will be spoken at R8. (Note: The clock rate will
remain at R8 from then on unless changed).
Example:IF IPl IR5 IS YOUR NUM IR8 BER IR4 FOUR FIVE IR9
NINE ?
In this example, we can make a question sound more like a
question by adding pitch changes at important points in the
sentence. "IS YOUR" and "NUM" are spoken at R5. "BER" is raised
in pitch to R8 and then, "FOUR FIVE" (you could also use 45) is
pronounced at a lower frequency of R4. Finally, "NINE" is
raised in pitch to R9 to end the sentence in a questioning
tone. The question mark will only be spoken if the punctuation
modes (IA or IM) are invoked.
Page 12

Note: When using the manual inflection mode f it is important to
set flat inflection (IF) mode or the algorithm will try to add
automatic inflection in addition to that manually selected.
Also, pitch and clock rates may be changed at any time in any
mode.
Punctuation modes
There are three modes for pronunciation of punctuation in
the MicroVox. The user options are:
IA (all mode —all punctuation pronounced)
1M (most mode — all punctuation pronounced except return,
linefeed, and space)
IS (some mode — only unusual punctuation pronounced)
When the MicroVox is turned on it is in "some" mode. In the
IM mode spaces between words are treated as pauses and can be
used to regulate the pace of speech or emphasize particular
words.
The MicroVox recognizes and pronounces all ASCII characters
with codes between hex 20 and hex 7F. The operating system does
not recognize control codes other than BACKSPACE (08) , TAB
(09), LINE FEED (0A), RETURN (0D), an ESCAPE (IB). Receipt of
other control codes or nulls, can have unpredictable results
since the MicroVox uses some of them for internal coding.
Illegal control codes should be avoided in the text sent to the
MicroVox.
On Line / Off Line Mode
The MicroVox can be selectively turned on and off line (it
has to remain powered, however). This capability allows it to
be attached in parallel with another peripheral such as a
printer, yet not speak what is being printed. The control code
is:
10 (On Line - MicroVox is operational. It responds to all
device codes and text input)
IQ (Quit - Off Line - MicroVox only responds to 10)
Default Modes
When the MicroVox is powered up certain default modes are in
force. They are equivalent to entering the following commands:
Page 13

!0 on line
IP1 IR5 low base pitch, clock rate #5
IF flat intonation
IT text to speech mode
IS some punctuation
IL Line by line pronunciation
ID9 wait for carriage return phrase terminator
(When shipped from the factory, MicroVox is set for 300 bps, 8
bit words, no parity, 2 stop bits, and software handshaking)
At any time these defaults are to be changed, simply send
the control code to the MicroVox. The codes can be transmitted
separately or imbedded in text. For example, entering THIS IS A
TEST, and a carriage return will result in that phrase being
spoken with no intonation. To add automatic intonation the
sentence becomes (all sentences are presumed to end with a
carriage return):
II THIS IS A TEST
From this point on all spoken text will have automatic
inflection unless flat intonation is resumed with IF.
As previously mentioned, intonation can be added selectively
or by the automatic algorithm. You can say the following
sentence four ways:
1. text to speech, no added inflection
IT IF
PLEASE ENTER YOUR ACCESS NUMBER
2. automatic inflection in text to speech mode
IT II
PLEASE ENTER YOUR ACCESS NUMBER
3. selected inflection in text to speech mode
IT IF IP1 IR5
PLEASE IR8 EN IR5 TER IR7 YOR IR5 ACCESS NUMBER
4. phoneme input mode with selected intonation
IF IC IP1IR5
P L El Y Z PAl PAl PAl PAl IR9 EHl EH3 N IR5 T ER PAl Y
IR8 02 02 02 IR5 R PAl IR7 AEl IR5 K S EHl EH3 S PAl N UH1
M B ER
These examples demonstrate various ways in which the user
can increase intelligibility of the synthesized speech. The
MicroVox is completely programmable, you can combine text to
speech with either selective or automatic intonation or
optimize pronunciation by choosing exactly the pitches and
phonemes you wish. An exaggerated example of combined pitch and
phoneme control can actually allow MicroVox to sing as
demonstrated in a bar of "happy birthday" and a musical scale.
Page 14

"Happy Birthday"
IC IP3 IR3
H H H AE1 AEl AEl AEl AE1 AEl P P IP2IR5 Y Y Y IP31R5 B ER
ER ER ER R TH TH TH TH 1R1 D Al Al Al A1 13
IR9 T IU IU IU IU U1 U1 U1 U1 Ul IR7 Y1 IU IU IU U1 U1 U1
U1 Ul Ul
!C
1P1 IR1 D D El El Y Y Y
I PI IR5 El El El Y Y Y
!P1 IRll EH1 EH1 EH1 EH2 F F P
1P2 1R5 D J J El El Y Y Y
1P2 IRll Al Al Al Al Al Y
IP2 IR14 B B El El Y Y Y
1P3 IRll S S El El Y Y Y
IP3 IR15 D D El El Y Y Y
Summary Table of Device Codes
10, IQ - On line and Off line
IK - synchronize speech and text
!L - line by line pronunciation
IW - whole text pronunciation
IE - each letter pronunciation
IC - pronounce by direct phoneme input
IN - produce musical notes
IT - pronounce by text-to-speech algorithm
!A, IM, or IS - speak all, most, or some punctuation
IF - set monotone or flat intonation
II - set automatic inflected intonation
IP and IR - set intonation base pitch and clock rate
ID1-1D8 and ID9 - set phrase terminator delay
SETTING THE SERIAL PORT
DTE/DCE Setting
Behind Jl (the DB-25 serial connector) on the PC board is a
2 by 3 header and two jumpers. These jumpers set whether pins 2
and 3 are transmit data and receive data respectively or vice
versa. As received from the factory, the jumpers are in the DCE
position and pin 2 is RD and pin 3 is TD. To reverse these
designations, place the jumpers in the DTE positions.
Data Rate
SW2 is the data rate (sometimes called BAUD rate)
selection switch. The data rates are listed along side SW2.
SW2 can be either a 2 by 8 or 9 position Berg type pin
connector or a 16 pin DIP switch. If a Berg connector is
installed, a jumper is provided to select the desired data
rate. Simply place it across the pair of terminals next to the
desired data rate.
Page 15

If SW2 is a DIP switch, close the switch position next to
the desired data rate. Only that one position should be closed
and the other seven positions should be in the open position.
For 75 bits per second, it will be necessary to attach a
physical jumper across JP1. All positions on SW2 should be left
open.
Handshaking
For software handshaking, switch position 3 on dip switch
SWl is set in the closed position. For hardware handshaking,
switch position 3 is left open.
If standard EIA RS-232C serial connections are used, the
sending computer hardware will detect and examine the RTS
signal and determine whether the MicroVox is ready to receive a
character or, if busy, take appropriate action.
With software handshaking, the MicroVox will send the
character " R " to the computer when it is unable to receive more
data, and will send "B w to the computer when it is again ready
to receive data. It is the responsibility of the computer
programmer to write the software necessary for the use of these
options.
Finally, it is possible to use the MicroVox with no
handshaking by simply invoking the software handshaking mode
and ignoring the handshaking transmissions. In this case, it is
the user's responsibility to insert timing delays in the
program so that data will not be sent to the MicroVox faster
than it can handle the data.
Word Length, Parity and Stop Bits
Three switch positions on SWl set the transmission protocol.
The following is a list of the eight possibilities and their
functions:
Function
Position 6
Position 7 Position 8
7
bits.
EP, 2 SB closed closed
closed
7 bits.
OP,
ZSB
closed
open
closed
7 bits. EP, 1SB closed
closed
open
7
bits,
OP,
1SB closed
open
open
8 bits. 2 SB open
closed
closed
8 bits.
1SB
open
open closed
8 bits. EP, 1SB open
closed
open
8 bits.
OP,
1SB open
open open
EP = Even Parity OP = Odd Parity SB = Stop Bit(s)
Page 16