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

The ViaVoice tts

Are you sure what you are seeing is an issue with the TTS? You will
get the bahavior you observe if you view a web page which contains
characters outside the normal ascii range, such as UTF-8 characters
used for line drawing or bullet points etc. 

If possible, I would find a page which reproduces the problem you have
observed and post it to the list and we can see if others observe the
same behavior. You may also be able to find a sighted friend to view
the page you are looking at and find out what is in the page at the
point where you are hearing the text you feel is 'spurious". Note that
what is displayed on the screen is somewhat dependent on 

    1. The font being used. If your font does not support the
       character, you will have just a square box. 

    2. Whether you have EMACS_UNIBYTE set or not. If it is set, you
       are likely to just see the octal representation of the

    3. The encoding settings for whatever web browser you are using
       i.e. w3 or w3m. 

    4. The language environment your emacs is using i.e. UTF-8 or
       ISO-8859-1 etc. 

Its interesting you have noted this only with ViaVoice as I've seen
this behavior with both the software dectalk and the hardware dectalk

You can also filter lines containing repeated characters and often
these are extended line drawing characters which when spoken say 

"a circumflex a circumflex acircumflex"

many times. To do this, you can use dtk-add-cleanup-pattern, which is
normally bound to C-e d a

I suspect the issue you are noticing is not TTS specific and if you
have been using flite or some other TTS interface, that interface
filters out characters of this type, which could be why you have not
noticed it before. While I do/have observed this behavior, its rare,
but I think I played around with my w3m settings quite a lot a while
back to get things sorted out and I have added to the default cleanup

It is possible there are improvements which could be made to emacspeak
in this area. However, this is a very unstable and uncertain area at
present. The next version of emacs is supposed to have much better
support for things like different character encodings and things like
multibyte character sets. However, even when emacs gets things sorted
out, its still going to be a problem for emacspeak because some TTS
engines only support basic ASCII characters and will crash if passed
multibyte characters or will try to interpret them as two characters

Having said all of this, I need to also point out that this is not an
area I have much knowledge in. Being an english speaker, I don't need
to worry about other character encodings too much. I'm running on
Debian, which has switched to a UTF-8 default encoding, which is what
I'm using at the OS level. To what extent this impacts on things,
especially interfaces to other programs like w3m, I don't know.
However, I'm pretty certain the issue you have noticed is not due to
spurious output being created by the TTS - its about how emacspeak is
interpreting these extended characters and what it sends to the TTS


Kalyan Mukherjea writes:
 > Hi, 
 > I have been trying to use the ViaVoice tts and coming up with a really
 > strange phenomenon which neither Tim  nor  Robert D. Crawford have
 > reported during the course of their discussions on this list.
 > I am running  a Fedora Core 3 system and have installed ViaVoice from
 > the rtk which Oralux plan to use for their Oralux Gold CD. I am
 > running VV using the OS simulation.
 > The problem is one of "spurious output".
 > Let me explain with a specific instance.
 > Trying to listen to Raman's blog on Voice locking and ACSS
 > after line 13:
 > ` 4. The numbers in voice-bolden-settings as an example:'
 > I hear "a circumflex inverted exlamation point" repeated several times and
 > the explanation following becomes impossible to figure out because of
 > this spuriae.
 >      Trying  out the voice sampler I can see how the voice locking can
 >      make for a much enhanced audio desktop but unless this current
 >      problem is resolved, it is not clear how useful this tts  will 
 >  b.
 > Any suggestions would be appreciated.
 > Cheers. 
 > Kalyan

To unsubscribe from the emacspeak list or change your address on the
emacspeak list send mail to "emacspeak-request@cs.vassar.edu" with a
subject of "unsubscribe" or "help"

Emacspeak Files | Subscribe | Unsubscribe | Search