🚀 go-pugleaf

This mail is in response to the recent inquiry about voice synthesis.
Since several people have an interest in this subject, I'm posting
this response (long).

I built 2 voice synthesis projects.  
	The first project I designed myself and interfaced to an Apple 
	IIc serial port (or any RS232 by changing connectors).
	It used the SPO256-AL2 speech synth. chip sold at Radio $hack
	(Cat. # 276-1784), and a General Instruments AY-3-1015 UART 
	clocked with a 555 timer running at 4.8 KHz.  The UART does 
	a serial to parallel conversion.  I modified the microcomputer
	interface diagram that came with the SPO256-AL2 to include the 
	UART, to not include the optional ROM, and cleaned up the use of 
	some strobes/handshakes.
	With this circuit you will be required to do some minimal
	programming on your PC to build words from allophones.
	Through programs, and manually, I sent character strings 
	(representing speech phonemes) out the Apple serial port to 
	the voice circuit and amazed my family and friends.  I sold 
	the Apple before ever doing any serious programming.
	After graduating speech synthesis kindergarden, I moved on
	to bigger and better things; text-to-speech.

	For my second project I purchased a circuit board from RFJ
	Engineering in Florida (305 323-9039).  This circuit uses
	the CTS256-AL2 text-to-speech chip (Radio Shack Cat. #276-1786)
	and the SPO-256-AL2 speech synth chip.  I've attached mail with
	more details on the circuit board.  NOTE- circuit board requires
	a slight modification to the power supply--see attached mail.
	With text-to-speech, an ascii word or phrase, such as "hello fred", 
	can be sent out the PC's serial port to the voice circuit, and 
	will be pronounced in English.  Neat!
	I recommend this project because it was simple, cheap, and I spent 
	more time on application and less time on design.  Here at Bell Labs 
	we use the "ctrm" communications software to log on UNIX machines 
	from PCs.  From UNIX with simple shell scripts, or from DOS with
	batch files, LOTS could be done here.  ("Good morning Mark... It is
	Wednesday March twenty-second... eight-o-four a.m... You have mail.")
	My scripts are available to anyone upon request.

Finally, after graduating from voice-synthesis jr. high, I moved on to the 
juicy stuff:

	My house is X-10 automated; lights, appliances, music, thermostat, etc.
	In addition to controlling these things from a wireless remote, any 
	telephone in the world, automatic scheduling by PC, and table-top
	control centers, I can do the following:

		Me:  	Godfrey,
		B.I.B:	May I help you?  
		Me:	Table lamp.
		B.I.B:	O.K.
		Me:	Off, please.
		B.I.B:	As you wish.
		(BIB's voice responses randomly differ every time)

	B.I.B stands for "Butler-In-A-Box" and is also a hands-free duplex
	speakerphone and PIR alarm.  Voice quality is superb.  A voice-
	controlled tricked-out house will run ~$2K.  And More fun than 8514
	and super-VGA!
	
Lastly, after scanning PC magazines for info on voice-mail, etc., I found
this demo phone number for WATSON voice-mail products:  1-800-6-WATSON 

Gladly I will share advice/info including my own schematics and programs
with any voice-synthesis hackers out there.

Mark Blumenthal
AT&T Bell Laboratories
att!ihlpb!markb
312 979-3676

(2 attachments)
__________________________________________________________________________
Date: Thu, 5 Feb 87 23:05:51 PST
From: cmcmanis@sun.uucp (Chuck McManis)
Subject: Text to Speech Chips
Organization: Sun Microsystems, Inc. - Mtn View, CA

A couple of people have mentioned that they are interested in using
the text to speech chips that Radio Shack sells. Well it turns out a
place called RFJ engineering in Florida sells a bare PC board that
does everything you need. They advertise in Computer Shopper. The
board has sockets for extra RAM (extended conversion buffer) and your
own rule exceptions EPROM. I built it and it worked right off. There
are level shifters for RS-232C operation but I have been using just
straight TTL serial on my robot. This board makes a great diagnostic
console on a mobile platform. 


 -Chuck McManis
uucp: {anywhere}!sun!cmcmanis   BIX: cmcmanis  ARPAnet: cmcmanis@sun.com
These opinions are my own and no one elses, but you knew that didn't you.
__________________________________________________________________________
>From sun!pepper!cmcmanis  Fri Apr 10 22:56:12 1987 remote from ihnp4
To: ihlpg!markb
Subject: Re:  Text to Speech

Mark the following is a followup message I posted to sci.electronics :

Ok, I mentioned earlier a board that does this, here are the particulars :

The company is called RFJ Engineering and their phone number is 
(305) 323-9039, their address is P.O. Box 4166, Sanford, FL, 32772.

They sell a bare board for $24.95 that uses the radio shack (actually
General Instrument) Text-to-speech and speech synthesis chips. This
board has an RS-232C port on one side and an amplifier/speaker on the
other. What ever ascii text goes in one side, comes out the other as
spoken words. It also has a parallel port input which you could connect
to a centronics port (with appropriate connectors of course). After
collecting the parts it took me about an hour to assemble, your mileage
may vary. I suggest that anyone who builds this *not* use the on board
5V regulator, and instead buy a radio shack wall bug power supply that
supplies +5, +x and -y (where X and Y are greater then 3, these drive the
RS-232 chips) and hook it up directly. 

Of course I have no financial interest in anyone mentioned, I built one
I use and it was fairly simple. It cost *a lot* less than $1000. More
like $50 including snacks and beverages.

P.S. It does *not* need an EPROM. The EPROM is for 'exceptions' and can be
     left off.

-- 
--Chuck McManis
uucp: {anywhere}!sun!cmcmanis   BIX: cmcmanis  ARPAnet: cmcmanis@sun.com
These opinions are my own and no one elses, but you knew that didn't you.