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

Re: Ensoniq capabilities

On 2/14/2013 2:52 PM, Scott Alfter wrote:

In article<alpine.LNX.2.02.1302141522110.15069@darkstar.example.org>,
Michael Black<et472@ncf.ca>  wrote:

On Thu, 14 Feb 2013, Steve Nickolas wrote:


On Thu, 14 Feb 2013, Michael J. Mahon wrote:


Phoneme synthesis with the Ensoniq would only require about 64 short
prerecorded phonemes and standard phoneme synthesis software--quite simple,
actually, but having the usual "robotic" speech quality.


I almost wonder if a GS could do something like Vocaloid or if that's pushing
the cpu too far.


Steve Ciarcia had a "Circuit Cellar" in Byte, I think it was 1978, where
he showed how to give a computer some speech output.  He did it the simple
way, sample the words, then play them back.  He was obviously using a
fairly low sampling rate.

The problem with that was storage space.  It was for an 8bit computer, so
RAM was limited to 64K.  I can't remember if he was using a floppy drive,
but even then that was limited compared to what's available today, 350K at
tops.


Way back in '85 or '86, I wrote a speaking math-drill program that ran on my
IIe as a Boy Scout project.  I used some audio recording/playback code
cribbed from Nibble.  Words were sampled one-at-a-time and saved to disk.
Since I didn't have a hard drive, I preloaded the samples into /RAM and
loaded them from there in sequence to get it to say what was needed. With
just numbers and a handful of operators needed, not much storage was needed.
It actually ran at a halfway-decent speed, even though it had to load words
one-at-a-time (so that "6*3=18" would load samples for "six," "times,"
"three," "equals," "eight," "teen" in sequence.)  Loading samples from RAM
disk kept the delay between samples to a minimum.

   _/_
  / v \ Scott Alfter (remove the obvious to send mail)
(IIGS( http://alfter.us/            Top-posting!
  \_^_/>What's the most annoying thing on Usenet?


In early 1981, I became intrigued with the idea of doing 1-bit sampling
of speech using the cassette input port.  1-bit sound is "infinitely
clipped", so it sounds like a cheap radio running with the volume too
high.  ;-)

The m/l routine to sample the cassette port at about 8kHz and pack
eight samples per byte are quite simple, as is the corresponding
playback routine.

I packaged all this as an Applesoft program and used it to read hex
dumps of Apple memory (so I could easily check typed-in listings).

I've put the program on my website, and if you list it, you'll see all
the "scaffolding" code used to capture the sampled digits, trim them to
the desired segments, and construct the DATA table of pointers into the
captured speech.

As I said, it's 1-bit sound, but it is quite understandable with some
practice.  And you'll understand why pilots say "niner" to make it more
distinct from "five".  ;-)

The Ensoniq chip can sample with 8-bit precision, which is much more
than is required for clear speech, though it's a little trickier to set
up to record and playback.

-michael

NadaNet 3.1 for Apple II parallel computing!
Home page:  http://home.comcast.net/~mjmahon/

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