                               CHAPTER 1.
                      GETTING TO KNOW DECTALK PC.
                                   
This  chapter provides a general description of the DECtalk PC speech
synthesizer  which incorporates the latest DECtalk speech  microcode.
It covers configuration, installation and operating mode.
The DECtalk PC Speech Synthesizer Option
DECtalk PC is a speech synthesizer option card intended for use in PC
based  systems. ASCII formatted text,  which is input to the  DECtalk
PC  via  the  PC  internal bus is converted into synthesized  audible
speech heard through an external speaker.

The complete option includes the following components:
       DECtalk PC  speech synthesizer option card,
       External Speaker with volume control and headphone jack,
       Getting Started Reference Card (print/braille),
       Installation Manual,
       Installation Guide Audio Cassette,
       DECtalk PC Software Diskettes (3-1/2"),
       Warranty Registration Card,

DECTALK PC OPTION CARD.
The DECtalk PC is a three quarter length 8 bit option card which plugs into an
IBM PC or 100% compatible internal bus. The bus may accept XT, ISA or
EISA (8, 16 or 32 bit) options.

The major components on the board are as follows:
       INTEL 80C186 16 bit Microprocessor,
       TMS32010 (Digital Signal Processor),
       1MB of DRAM,
       Digital to Analog Converter (DAC),
       Audio amplifier with anti-aliasing,
       Output Jack (1/8" mono),
       8 bit IBM PC bus interface.

DECTALK PC SOFTWARE.
The  microcode for the DECtalk PC option card, the DOS driver and the
installation program are supplied on floppy diskettes. Providing  the
DECtalk  microcode  on  diskettes  provides  the  ability  to  easily
distribute upgrades to the synthesizer functionality.

SOFTWARE INSTALLATION.
The  installation program will prompt the installer, via  both  tones
and  screen  menus,  through the install procedure.  The  instalation
procedure  will load the software onto the system's hard  disk,  load
the  DOS  driver and will initially load the board with its software.
Successful  installation will cause the board to  speak  its  startup
message.

The  install program will also, as a default, update the AUTOEXEC.BAT
file so that the microcode will be automatically downloaded from  the
hard disk to the DECtalk PC upon system startup.

AUTOEXEC.BAT  INITIALIZATION COMMANDS.
The  commands  added  to the AUTOEXEC.BAT file  for  the  DECtalk  PC
perform the following functions:

        DT_CONF looks for DECTalk PC option cards installed in the PC
bus.  If it finds any, it will attempt to initialize them and reports
the status.

        DT_DRIV  loads  the DECtalk PC TSR into the PC's  memory.  It
will  not load if DT_CONF was not able to successfully initialize  at
least one DECtalk PC module. This TSR can be loaded into upper memory
(LOADHIGH).

        DT_LOAD loads the 1MB of memory on the DECtalk PC option card
with  all  the  software modules required to implement  the  text-to-
speech synthesis function. Application specific dictionaries can also
be loaded.

       DT_START causes the DECtalk PC to speak it startup message.
POWER-UP SELF-TEST

        Upon successful module initialization and software load,  the
DECtalk will say:

       "DECtalk Version 4.x is running."

OPERATING MODE.
        The  DECtalk  module is ready to operate when the  successful
power-up  message  is spoken.  In this state, the  DECtalk  board  is
controlled via the DECtalk TSR driver, DT_DRIV.EXE.
What's New In DECtalk PC

        Many  users  may  be  familiar with the  serial-line  DECtalk
speech  synthesizer.  The DECtalk PC with DECtalk  V4.x  software  is
compatible  with the linguistic capabilities and high quality  speech
of  other  DECtalks. However, there are a number of  differences  and
improvements.

LATEST VERSION OF SPEECH SYNTHESIS
        DECtalk  software  contains the  latest  version  of  DECtalk
speech  synthesis.  This incorporates a number  of  improvements  and
modifications  to  the  firmware used in earlier  ROM-based  systems.

These include:
     High Quality Speech
        DECtalk's  speech retains its high quality.  In  addition,  a
number  of  improvements  have  been made  in  functionality,  speech
quality and naturalness.

     Increased Word Pronunciation Accuracy
        The  accuracy  of word pronunciation is higher  than  in  any
previous  version of DECtalk. Words are seldom mispronounced.   There
have  been  significant improvements in the accuracy and  quality  of
letter-to-phoneme rules. Also, DECtalk contains a very large built-in
dictionary  which assists both the pronunciation of individual  words
as  well as its rhythmic naturalness. The fixed dictionary in DECtalk
software  is  many times larger than any previous DECtalk  dictionary
and  the  entries  are more complex and contain a  wider  variety  of
information  to  increase naturalness. As in the past,  the  built-in
dictionary  is inaccessible to the user although the user  dictionary
is accessible and modifiable.

     Improved Pronunciation Heuristics.
         Certain   heuristics  have  been  improved  and  made   more
intelligent.  For  example, DECtalk is able to better  recognize  and
parse  unpronounceable sequences such an uppercase initialisms  (FBI,
AAA,  etc.) in addition to the normal unpronounceable sequences  such
as those with no vowels (CBS, NBC).

     Larger User Dictionary.
        DECtalk PC contains a user dictionary. This dictionary can be
used to load application-specific words, DOS-specific terms, and  the
like.   This dictionary is now much larger than those of the  earlier
DECtalks.  The size is variable and depends upon what other  software
is  resident on the board.  However, the usable space may be as  high
as  350K  bytes and would be sufficient to load thousands  of  words.
Because  of  the  large  dictionary, developers can  now  input  many
keyboard  key names and commonly used DOStm and PC application  words
and commands (e.g., autoexec.bat, config.sys, etc.).

Note:  Dictionaries  are  searched in the order  in  which  they  are
downloaded.

     New Name Pronunciation Capabilities.
        One  of  the  items most requested from previous versions  of
DECtalk  has been an improved ability to pronounce proper names  such
as first names, last names and street names.  DECtalk is now capable,
for the first time, of pronouncing proper names with a high degree of
accuracy and greater level of intelligence. The rules behind the name
pronounciation   routines  were  originally   developed   for   large
commercial telecom applications but have now been modified to run  on
the  DECtalk.   This can be run in different modes and  should  be  a
great  help to applications which require peoples names and addresses
to be pronounced correctly.

     Tone Generation Capability.
        DECtalk  will generate certain tones (e.g., for margin  bell,
alert  etc.)  in  addition to speech sounds. It  also  maintains  its
ability  to  sing  and  the developer still has  the  flexibility  of
modifying acoustic parameters such as pitch, duration and the like to
create different voice qualities.

     Settable Volume Control.
        DECtalk contains settable volume control both in hardware and
in software. There is a volume control on the loudspeaker which comes
with  the  PC-board  and  there is also a  volume  control  which  is
settable by a command sequence.

     Faster Speech Rate.
        Speech rate on DECtalk runs from a slow speed of 75 wpm to an
upper  rate of 650 wpm. This is as much as 300 wpm faster than  other
DECtalks  and  thus  is more useful for applications  where  scanning
large bodies of text is necessary.

     Letter Mode, Word Mode and Clause Mode.
        DECtalk  is  able  to  immediately  speak  single  characters
without  waiting for an entire clause to be buffered. This is  useful
in  applications requiring tactile feedback for what was typed on the
keyboard. It also provides normal clause buffering for highly natural
speech.   DECtalk   can  speak  letters,  words,  phrases,   clauses,
paragraphs and whole documents.

     Immediate Stop Speaking.
         DECtalk   allows   the  application  to   terminate   speech
immediately  instead of waiting for the buffered  text  to  complete.
Stop  Speaking can be accomplished by commands such as [:pause] which
acts  immediately. DECtalk will still accept commands.  DECtalk  also
has  the  ability to resume speaking where the text last  off  or  to
flush text and immediately start speaking new text.

     Shorter Command Strings.
Many of the command strings such as change rate, change voice, start,
stop,   index,  and  index  reply  may  be  shortened   to   simplify
applications.

     Increased Buffer Size.
Buffer  size in DECtalk has been significantly increased.  The  input
buffer size is 4Kbytes and the output buffer size is 4Kbytes.

ADDITIONAL MODIFICATIONS IN V4.2B
        V4.2B is the latest version of the DECtalk speech synthesizer.
A  number of modifications have been made and new functionality   has
been  added.  These include new TSR calls and user  commands,  multi-
module support and MS-Windowstm support.

SPEECH FUNCTIONALITY IMPROVEMENTS.
There  have  been a number of improvements to the speech quality  and
functionality of DECtalk in V4.2B. Speech improvements are an on-going
effort  and  each  successive version of  the  software  should  have
improved  functionality  as  well  as  improved  intelligibility  and
naturalness.

Citation Mode.
A  Citation  Mode  has  been added for certain  function  words  which
needed to be pronounced differently when used in isolation as in  some
of  the  educational  and literacy applications.  When  this  mode  is
invoked  the word to is pronounced [t'uw], the is pronounced  [dh'ah],
for  is  pronounced [f'or] and so on.  The command to invoke  citation
mode is [:mode citation on].  The default mode is ON.

Pronounce Name Expanded.
        In  V4.0,  the  [:pron name] command could  be  used  at  the
beginning of a character string only. This has been expanded so  that
it  will  work  in any syntactic context. Therefore,  if  a  name  is
mispronounced,  or  a  name  field  within  the  application  can  be
identified,  you can now invoke the [:pron name] command anywhere  in
the text, e.g., The customer name is Joseph [:pron name] Carbone.

Punctuation Pronunciation.
        The  command  [:punc some] works in a more  intelligent  way:
"period"  is  no  longer  pronounced in any context,  e.g.,  ellipsis
marks.  Spell  mode will no  longer activate except  in  alphanumeric
strings.  [:punc  none]  now  processes  (but   does  not  pronounce)
punctuation  so  that ":", ";", "-" and the like  will  force  pauses
but will not be spoken.

Spelling Speed.
        Spelling  speed is tied to speaking speed. No  special  rules
will   be  used  for  spelling more slowly or quickly  than  the  set
speaking  speed. However, the speaking rate can  be easily  increased
or decreased in the normal way.

Automatic Homograph Pronunciation.
In  DECtalk  V4.1 homographs (words which are spelled  the  same  but
pronounced  differently) are chosen automatically thus relieving  the
user of having to decide on the phonetics.

For example, if DECtalk speaks the following pair of sentences:

       I'll refuse the refuse.
       They'll produce the produce.

You  will notice that the correct form is chosen automatically.   Two
other  commands for homographs have been added. These are [:pronounce
primary]   which   provides  the  primary  (usually  more   frequent)
pronunciation,   and  [:pronounce  alternate]  which   provides   the
alternate  pronunciation. Therefore, [:pron prim] august  will  yield
['aogaxst]  'the 8th month of the year,' and [:pron alt] august  will
yield [aog'ahst] 'serious, somber.'

DTDEMO SOURCE FILE.
The  source files for DTDEMO are now included for reference  purposes
for programmers and developers.

FAST TYPING MODE.
      To  allow typing mode (for audible feedback) to work as fast  as
possible,  a  new TSR call (FLUSH_CHAR) has been added.  (see  below).
This is used in conjunction with a keyboard mapping table which can be
customized  with  a new [:TYPE DD ASKY] command. Also,  the  pitch  of
selected  ASCII characters in the table, such as CAPS, can be modified
using a new command [:pitch dd] where "dd" is  a delta pitch change in
Hertz.

NEW TSR CALLS AND USER COMMANDS.

Rapid Flush-and-Speak-Key.
      FLUSH_CHAR  (0x2a) provides a rapid flush-and-speak-key function
e.g.  "typing  mode",. When invoked it flushes any pending  speech,  a
built  in translation table is used to convert characters to the sound
equivalence  and  the  results are sent directly  to  the  synthesizer
itself,  bypassing  significant pieces  of  the  DECtalk  code.  .  In
addition, a means of assigning pitch to a character (such as an upper-
case  character) has been provided. Also, the typing table itself  may
be  modified  so  that  keys can be redefined. A  global  delta  pitch
command  is  provided  to  set the change in  pitch  for  emphasis  of
uppercase  characters  and the like. Its  default  is  a  35  a  hertz
increment.

Priority Change.
        IO_PRIORITY (0x26) sets the comm task inside DECtalk  to  the
highest  priority. This results in much faster indexing which  should
be the preferred setting in most cases.

Indexing Function.
        LAST_INDEX (0x25) returns the last seen index as an  int   in
the  C  register.  It  returns a retry if it has  not  processed  any
indexes since the last request.

Digitized Speech Output.
        V4.2B allows for the playback of digitized sound files which
are in the Sound Blaster VOC format.  8-bit mono digitized sound files 
recorded with a sampling rate of 4Khz to 22KHz Digitized sound files 
can now be played back through the DECtalk  PC.

Note:  Digitized sound is unidirectional on  DECtalk. Sound files
cannot  be  created using the board but they can be played  back.  To
create files, another device must be used.

The  new  TSR  commands for implementing the digitized  mode  are  as
follows:

DIGITIZED_MODE  (0x18).
Puts DECtalk into digitized playback mode.

TEXT_MODE (0x19).
Puts DECtalk into text-to-speech mode.

DIGITIZED_DATA  (0x1a).
Sends digitized data to the DECtalk.
There  are also a number of new immediate voice change commands  also
listed in Chapter 3.

MULTI-BOARD SUPPORT.
In V4.0 of DECtalk, only one board could be supported per PC system..
In  V4.2B,  a multi-threaded support has been implemented and four  (4)
DECtalk PC boards can now co-exist in one system. This will be useful
in  the development of multi-line telephone-based applications.   The
limitation  on  the number of boards is based on (a)  the  number  of
available  slots in the PC (up to 4); and (b) the number of available
I/O  addresses.   The switchpack on the DECtalk  PC  allows  for  two
switches (4 positions) for I/O addresses. Thus, the maximum number of
boards currently allowed is 4.  The lowest numbered I/O address (240)
is  module  0.  The  next lowest is module  1,  and  so  on.  In  the
multiboard  mode,  register SI indicates  which  board  you  wish  to
communicate  with on all calls and it must be specified if  there  is
more  than one board installed. If there is exactly one board in  the
system,  SI  is  ignored. To determine the number of  boards  in  the
system, use the GET_STATUS call which returns the number of installed
boards in the SI.

Note: Please refer to appendix F.

WINDOWS SUPPORT
        An  MS-Windows DLL (Dynamically Linked Library) is  available
for  DECtalk.  To request more information call the DECtalk  Bulletin
Board at  508-467-3806 and leave a note.

End of Chapter 1.
                 
