tedk@ihuxv.ATT.COM (Kekatos) (09/14/88)
There have been several discussions on sci.electronics about
speech chips. Here is an article that I wrote a while ago.
------------------
Review and Information on B.G Micro Text to Speech Board.
COMPUTALKER SPEECH SYNTHESIZER
from B.G. MICRO (214) 271-5546 (Dallas TX)
I purchased the B.G. MICRO COMPUTALKER board for $89.95 (price is lower
now). It can be inserted into a PC, or operated in a stand-alone mode.
I have it set-up in stand-alone operation. I have it
propped-up between books on my book shelf. I built my own +5,+12,-12 power
supply that doubles as a book-end. A power supply is available from B. G.
Micro. I use a ribbon modem cable to connect it to one of several
different computers.
This Board uses the General Instruments SPO256-AL2 and CTS256-AL2
chip set. These are the same chips that Radio Shack sells. These chips
have been mentioned in several books on voice synthesis ("Chip Talk").
The board uses one slot (for power only) and requires one serial COM port
connection. The board can except ASCII TEXT from any computer (or terminal)
with RS-232 port. There is a 1700 byte input buffer with hardware handshake
signals. The board is shipped with a (IBMPC) Demonstration Disk, documentation,
and Schematics. The board has a speaker and a RCA type phono jack for external
speaker.
The serial port on the board can be set to one of seven baud rates up to
9600 baud. The COMPUTALKER begins speaking the received TEXT after it receives
a CR (0x0d) character.
The TEXT to SPEECH program within the COMPUTALKER is limited but is fairly
good for the price. It has trouble with speaking numerical text. Everything
must be spelled out. Such as "$104" must be sent to the COMPUTALKER
as "1 hundred and 4 dollars".
The voice quality is good compared to some voice synthesizers. It sounds
a bit "nasal", and take some time to get use to. It is best if there are
spaces between words and difficult to understand words are
spelled "phonetically". This take some experimenting.
I have read material on phonetic speech with PHONEMES and on using/constructing
ALLOPHONE speech with the SPO256-AL2, but I haven't found a way to by-pass the
text to speech conversion program. From my reading, the Votracs(tm) system has
control character sequences to alter tonal voice qualities and set control
modes. This is something like what a good printer would allow. I spoke with
the guy that designed the COMPUTALKER for B. G. MICRO, He says that there are
NO special control sequences that the voice synthesizer responds to, such to
alter the modes of operation or alter the speaking tone, etc.
----------
Ted G. Kekatos
backbone!ihnp4!ihuxv!tedk (312) 979-0804
AT&T Bell Laboratories, Indian Hill South, IX-1F-460
Naperville & Wheaton Roads - Naperville, Illinois. 60566 USAtedk@ihuxv.ATT.COM (Kekatos) (10/10/88)
In article <720@m10ux.UUCP> rgr@m10ux.UUCP (Duke Robillard) writes: |In article <2846@ihuxv.ATT.COM> tedk@ihuxv.UUCP (55624-Kekatos,T.G.) writes: |>Review and Information on B.G Micro Text to Speech Board. |> COMPUTALKER SPEECH SYNTHESIZER |>from B.G. MICRO (214) 271-5546 (Dallas TX) | | There's also a speech synthesizer for IBM-ish PC's from a company |called COVOX. 675 Conger Street, Eugene, Oregon, 97402 (503) 342-1271. |It's about $70. It consists of a speaker/amp box, about 4 inches high, |which uses a 9 volt battery and a little box (with the D/A chip) that |plugs into the parallel port. It's called the Speech Thing. comes with |device drivers and some examples in basic and C (mostly basic) and a |program to turn ASCII text into speech. It's pretty cool. | | Covox also sells a A/D speech converter. This consists of a microphone |and a 1/2 size board. It's suppose to generate input for the Speech |Thing mentioned above. This costs some more (about $90) and I don't have |it yet so I can't say if it's cool or not. | || Duke Robillard UUCP: {backbone!}att!m10ux!rgr There is one thing that needs to be understood. The COMPUTALKER board is a stand-alone hardware solution. The only thing that one needs to do is pass ASCII english text to the COMPUTALKER via a serial port. The board does the rest. I have printed material from COVOX on their products that are mentioned above. There product is only a D/A converter with a speaker. Your computer must do the rest. All the text-to-speech conversion be processed by *your* computer. If you have several different computers (as I have), you must have a version of the COVOX software for each computer. Granted, the COVOX solution IS more flexible, allowing the user to record/playback ANY sound(s). But there is a software dependency that could be a problem to some (non-standard) systems. Personally, I own two computer systems (Victor-9000) that have a CODEC chip on the motherboard. (I also have the Victor option board that allows recording.) A CODEC chip is a COder-DECoder. It is something like a A/D - D/A. Using *software* one can record and playback ANY sound from audio input. I have a small application that allows one to record "samples" of audio (up to 4.5 secords), each sample on a key on the keyboard. Then in playback mode, the keys can to pressed in any order to playback the "samples" (operating something like a sampling keyboard). Surely, This could also be done with the COVOX products. My point is... That for speech synth only, I feel that one of the many hardware solutions, such as the COMPUTALKER or a VOTRAX box, is a better value. (In the above context, when I say "hardware solution", I mean some type of dedicated system that performs the processing work.) This is only my own opinion. Ted G. Kekatos UUCP: ..!att!ihuxv!tedk (312) 979-0804 AT&T Bell Laboratories, Indian Hill South, IX-1F-460 Naperville & Wheaton Roads - Naperville, Illinois. 60566 USA