This is a read-only archive of the old Scratch 1.x Forums.
Try searching the current Scratch discussion forums.

#1 2009-09-05 14:18:00

realSAB
Scratcher
Registered: 2009-08-08
Posts: 6

ScratchTalker - Text To Speech

I played around yesterday and I think found a pretty slick way of playing a series of phenoms to have it sound "natural" (as possible with Scratch).  Having tried the
[ play sound until end ] block, which broke up the flow of the sound of the word.
I used the [ play sound ] in combination with a [ wait ] block and was able to "cut"
off the previous sound and play the next sound with a much better flow.  But, it has
a certain drawback.  The Phenom script has grown to a large enough proportion to make editing the waits or adding new sounds extremely slow.  I have tried to branch off using just the numbers in one lookup script, but Scratch doesn't seem to understand "If number is >= (zero) and number is <= (9)" too well.  I tried, but I refuse to have a string of "IF = 0 or IF = 1 or IF = 2.....or IF = 9" block just to branch to another script.

Any suggestions?   Of course, there is the "IF = 0, broadcast goto0" method.  And have a separate script for each sound.

Right now... The word is scanned for up to 4 phenoms.  For example: Kathee
would be K+wait, a+wait, th+wait, ee+wait.  Aside from organizing the tables
in a most common letter arrangement.  I can see no escape from having at least 60+
entries to cover the most common English sounds.  Right now the phenom script has about 30 entries and takes about 10 seconds to respond to any change.  I have been
"breaking" the script apart to edit it, and then putting it back into place.

Sub-scanning includes various offshoots from a base letter.  For example: o (oh)
currently has - o (short vowel), oh (long vowel), ow (blend), oo (double)
and or (as in for) sounds.   Trying to find a fast workaround for this one....

Thanks,
SAB

Offline

 

Board footer