1) If you are using words entered by a user (less predictable), you could
use the Speech Manager Dictionary routines? This is similar in practice to
the word substitution method except that it is done at the OS level so
should be theoretically faster?

2) If the words are predictable, why not enter them phonetically?

3) The stuttering effect is possibly the virtual memory? The speech routines
I use (Speech Primer at http://www.ScorpionResearch.com/futurebasic/) do not
stutter with real memory but sometime struggle if VM is on. If this is the
case, preloading may not help because the VM will still move the memory to
disk and the VM reloading exercise will probably result in no particular
performance improvement? I may be open to correction!

Good luck,

Deep











> From: Phil Yates <phil.yates@...>
> Reply-To: futurebasic@...
> Date: Sat, 13 Nov 1999 22:09:47 +0000
> To: <futurebasic@...>
> Subject: [FB] MacinTalk
> 
> Two problems.
> 
> Problem 1 :
> 
> I have a text string, something like :
> 
> "Speedbird 123 is airborne from Heathrow, Heading 270 Degrees, Over."
> 
> Now, Macintalk, although generally very good, has minor problems with some
> words, and I want to substitute phonetic words for the words it can't
> handle.
> 
> So if I have a Text string - say Temp$ - I want to check if it contains
> words I know it doesn't like, and substitute phonetics.
> 
> Let's say that I have 50 "iffy" words - can anyone suggest a FAST procedure
> whereby I can check Temp$, and substitute as necessary. Or do I have to do
> an INSTR with each iffy word, and then laboriously switch it out ?
> 
> Problem 2 :
> 
> I have so much going on that MacinTalk seems to load and then unload itself
> every time I want to say something - so that often, the first utterance is
> stuttered. It must be possible to load all the MacinTalk Resources into RAM,
> but which resources should I copy and paste into my application, and make
> them preloadable ? Any experts out there ?
> 
> TIA,
> 
> Phil.
> 
> 
> --
> To unsubscribe, send ANY message to <futurebasic-unsubscribe@...>
>