Heteronym suggestion

[McFortner]

One thing I've noticed is that heteronyms can give TA some problems. Depending on where in the sentence and/or if it is capitalized will result in different pronunciations. Could there be some way that TA would flag these words in a different color text so you could check and make adjustments so it would read right?

Wikipedia has a list of examples at http://en.wikipedia.org/wiki/Heteronym_(linguistics) that show's what I mean. The big ones I have noticed so far are nice/Nice, read, minute, and live. I bet others can elaborate on this from their experiences.

Just an idea.

Michael

[kdwhite]

neat idea. We'll take a look at it after 3.0 release.

[RicochetRita]

I have a whole list of these which I have to manually check/change when creating my own mp3s. The color idea would greatly speed up the process.

My lists (& how I currently force a particular pronunciation):
secreted --> seekreted
bow --> bough
live --> liv
live --> lyve
wind --> wynd
wound --> wow'nd
minute --> minit
minute --> my-nute
read <--> red
lead <--> led
tear <--> teer
lied <--> lide

These are pronounced differently as Nouns (accent on 1st syllable) & Verbs (accent on 2nd). I've had to create two dictionary entries for each--verbs get the odd mid-capitalization form:
prezent <--> pre-sent
project <--> proJect
suspect <--> suSpect
subject <--> subJect
object <--> obJect (also catches: obJective, etc)
content <--> conTent
console <--> conSole
contract <--> conTract
record <--> reCord

[geneven]

It's a good idea to allow these things to be corrected.

In addition to seeing these in a different color, I would prefer a different program that pops up each possibility with a few lines of context. Then the choices would be presented, with some explanation of what the choice means. You would run the program before having a text read to you, thus getting rid of these problems in advance.

I would say that this one problem is the most frequently cited horrible thing about text-to-speech. If you could solve this, the irrational opposition to text-to-speech would probably go away to a large extent.

If as I suggest this be made a different program, you could offer it before TextAloud 4.0.