[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Speech Input/Output on Apple II

xorxif wrote:

aiiadict@gmail.com wrote:


S.A.M. "Software Automated Mouth" is speech output for the apple II.
It requires no sound card (it can output through the speaker).  It is
the only software I have seen before today that would do this.

Echo, Mockingboard, DoubleTalk, Slotbuster were all hardware cards that
would output speech.  There was a card for the SAM software that would
improve speech output (d/a converter on a card)

Other voice software:
ghostbusters had digitized voice
Bob Bishop did some fun things with digitized voice...
--B.B. Talking Calculator
--B.B. Apple Lis'ner (speech recognition through the cassette port, all
through software)
--B.B. AppleVisions book, sound record and playback through cassette
port
--B.B. AppleVisions book, mentions speech recognition in last chapter,
through cass. port.


Voice input for the Apple II:
Covox Voice Master:  Connects to game port, recognizes 32 words (more
if you write your own code to get around the included software
limitations)
Voice Input Module: Speech recognition on a card + automated keyboard
Heuristics H2000: Speech recognition on a card
Scott Instruments Shadow/VET: Speech Recognition on a card


"The Voice" by Muse Software?
according to Bill Garber:
"Everything I'm getting from my searches
tells me "The Voice" also does voice recognition."

I've never heard of this one, but apparently "Lis'ner" by Bob Bishop
does the same thing.

Rich



Don't forget Sweet Talker, by Micromint.
I think it used some kind of add on card as well, made before Mockingboard.


Micromint also designed a speech recognition (and synthesis) card
referred to earlier, called the LIS'NER 1000.  It used an LPC
analysis chip driven by a microphone input to produce a stream
of linear predictive coding coefficients that could be correlated
with "training" samples to do recognition.  The samples were stored
in Apple memory, and an Apple program did the recognition.

Recognition accuracy was "moderate", but much better than any
scheme based on the 1-bit cassette input.  It was recommended
that no "dictionary" contain more than about 8 words to choose
from at a time.  More words were supported by going down levels
of dictionaries.

Only the Heuristics and Scott cards did better, because
they were computers-on-a-card, with well-honed algorithms
matched to their capabilities--and even they were primitive
by today's standards.

All were speaker-dependent, isolated word, quiet environment
recognition systems.  None did what most people wanted, which
was to listen to a room full of people with music playing and
respond to "computer, dim lights".  Come to think of it, we
still can't do that...  ;-)

-michael

Music synthesis for 8-bit Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."