the mask ROM. The pre-programmed vocabulary addresses
are listed for the mask ROM on the last page. Even without
the ROM, it should be possible using an embedded
processor (like an Arduino or a Raspberry Pi) to program
the synthesizer chip to output a series of allophones
(phonemes, or basic sounds used in English) to form words
that aren’t programmed in the ROM.
There is more documentation on the Wikipedia page
for this family of chips at https://en.wikipedia.org/wiki/
Looking around on the Web, it seems that several
people have gotten these SP0256 devices to work with
an Arduino. There is an Instructable that shows how to
do it. This one doesn’t use the ROM, but instead directly
excites a series of allophones. You can see it at www.
There is some code there for the Arduino IDE (integrated
development environment) that will get you started.
Looking at the library, it probably involved a lot of
typing to get all of the allophone codes right. There is also
an allophone address table document, as well as examples
of how to combine them to make common words. This is
what you’ll need to create first words and then sentences.
Unfortunately, it appears as though this design doesn’t
include any amplification of the output audio. It won’t be
very loud without doing something, so you might consider
an amplifier chip or building a simple one transistor class A
amplifier to make it louder.
Another example of using this chip can be found at
www.thefrankes.com/wp/?p=2490. It includes an amplifier
for the audio output — at least in the prototype. There’s
a video there too, as well as a printed circuit board. The
board doesn’t appear to include the amplifier, but there’s
no schematic so it’s not clear. If you watch and listen to the
video, you’ll see that this isn’t really the state-of-the-art in
speech synthesizers, but I think it would be a fun project
getting it to work.
Image licenses: Figures 4 and 5 are licensed under
Creative Commons 3.0 and were created by Walter Dvorak.
They can be found at https://en.wikipedia.org/wiki/
Kristen McIntyre is currently a senior software engineer at Apple working on operating systems. She recently came back from being
an entrepreneur in Japan. Previously, she
was a researcher at Sun Microsystems
Laboratories where she was researching
robustness and emergent properties of
large distributed computer systems.
Her career has spanned many diverse
areas. She started in the early ‘80s
designing high power linear amplifiers
and then spent about five years in Japan
architecting and designing precision
analog test systems, as well as learning
the Japanese language and culture.
She speaks, reads, and writes Japanese
Upon returning to the states, Kristen
joined Adobe Systems and became
one of the architects of PostScript Level
2 and its RTOS underpinnings as well
as the principal architect of AppleTalk
networking for PostScript printers. In
the early ‘90s, she became a consultant
and later founded an Internet service
provider and network consulting firm.
In 1999, Kristen decided to hang up her
entrepreneur’s hat and landed at Sun,
tried the startup thing again only to land
at Apple, where she’s been since. Kristen
holds a BS EECS from the Massachusetts
Institute of Technology.
Kristen has been interested in radio
since she was about five years old. When
she was small, she built many radio kits
including her favorite: the one tube radio
kit. She started in amateur radio around
1979 while she was at MIT by getting
her technician’s license. She built a 2m
repeater with an autopatch to use while
on campus at MIT. Kristen is American
Radio Relay League Technical Coordinator
for the East Bay Section and is President of
the Palo Alto Amateur Radio Association.
She gives talks all around California and
the Southwest on various technical topics.
She is licensed with the amateur radio
callsign K6WX in the United States, and
JI1IZZ in Japan.
October 2016 13