[Asterisk-Users] Asterisk bounty: email TTS

[Adam Goryachev]

On Wed, 2005-07-20 at 14:42 +0300, Tzafrir Cohen wrote:
> On Wed, Jul 20, 2005 at 07:01:48PM +0800, Craig Guy wrote:
> > How do you handle:
> >     RTF
> Not very common

Isn't this easily converted to text?? I thought the format for this was
pretty simple, but I could be wrong...

> >     Disclaimers
> >     Signatures (inc ascii art sigs)
> >     Virus scanner tags
> Any idea how to identify them to be ignored?

You 'should' be able to ID signatures (which should include
disclaimers/etc) by a -- at the beginning of a line by itself (or
something like that, there is some sort of standard for that). However,
most people probably don't know the standard, and so don't use it...

So, I think the best way to handle this (where it can't be detected) is
to allow the end user to press a key for at least these options:
3 Next message
6 Next line
9 Next paragraph

With the obvious opposites as well... So when you hear the beginning of
some stupid disclaimer, just skip to either next paragraph or next
message...

> >     mispellings
> no idea

Well, like some word processing applications, you can probably use some
dictionary based on common mis-spellings. If all else fails, just say it
as-is... tts isn't going to sound perfect anyway ... :)

> >     abbreviations and SMS type speech (CU L8R, ROFL), smilies
> >     quoted replies and nested quoted replies.
> Different intonation if at all

Interestingly... CU L8R would probably sound better than "see you later"
because it is based on the sound of those letters... I think I noticed
this somewhat with it when I was trying out some stuff.... sure, ROFL
wouldn't work, but again, this could come from the above dictionary,
and, you could also convert all uppercase words to be read as the single
letters rather than a word...
 
> > Sure you can do filters to handle some stuff but it seems to me to be a
> > never ending battle to keep filters current.
> The sender, subject and time are still readable, regardless of contents.

IMHO, it would be a matter of dealing with things as they come up. Like
wine (the non-emulator for windows) the first 90% is easy, it's the
other 10% that will take 90% of the time....

Yet 90% useable is a lot better than not at all.

BTW, I wonder if there is some application already written for email
which is designed for use with braille terminals or other limited output
style devices. 

Also, ever consider allowing the user to follow a hyperlink, and then
you can 'just' build a tts web browser as well :)

Just my 0.02c worth...

Regards,
Adam