Re: [orca-list] Orca: how to change voice dynamically

[Michael Whapples <mwhapples aim com>]

Before I start, I am just getting to know the orca code in detail myself so may not be able to point at exactly where changes should be made.

Reading what you say about how you might detect the language this should work on any unicode string. In this case could this be done only in the speech output stage? By this I mean Orca scripts produce speech output through the orca.speech module. Even if that isn't exactly where the changes would need to be made in the output process, you should be able to follow the path of the text to be output from there. You probably would also want to modify the Orca preferences offered to users so they could configure such a feature, but I am unsure where those changes would be.

Now I will give a few warnings on what you suggest (these are based on thoughts rather than trying what you suggest).

* If all output speech were to be tested for language, how will this impact on the responsiveness of orca?
* Users may want to turn it off (possibly for performance reasons, I use English and more than 99% of the time I am reading English with orca).
* Considering the possible amount of use, is there any particular applications where you would want such a feature? In my case about all of the text in other languages (the little of it I read) is from firefox.
* If there are specific applications, are there better ways of detecting language? I think openoffice knows about the language of the text (don't know whether or how we get that information).
* Might it be better as an application specific script?
* To remove the need for automatic detection (which may not be accurate in all cases anyway) would having a shortcut key for a language selection menu be enough?

I will be honest and say I am probably biased as my use of other languages is very low, but I am just cautious of this having a negative impact on a significant number of users if done wrongly.

Michael Whapples
On 02/08/2010 04:06 AM, anila susan wrote:

Hi Michael,

Thanks for the info.
I think we have an option to figure out the language of the text. We can use the unicode numbers to identify the language.

But I need to know, which code module or python script, reads the text. I need to make changes in that file. Also which script chooses the voice. If anyone knows these code details, please let me know.

Thanks a lot,
Anila

On Sun, Feb 7, 2010 at 5:49 PM, Michael Whapples <mwhapples aim com> wrote:

Hello,
I must admit I am not an expert on orca's code but here is what I guess might need to be done.

* You will need to detect the language of the text some how. This might be easier in certain applications as they may know this information but could be difficult in others (eg. a plain text editor such as gedit won't have language information).
* You will need some way for selecting the voice for each language (eg. I may have multiple english and german voices, how might it know which to use for german). This probably means adding settings to the orca settings window.

I imagine this might need changes through quite a bit of orca.

Michael Whapples

On 01/-10/-28163 08:59 PM, anila susan wrote:

Hi,

I'm  working on a text to speech synthesis using festival, for Indian languages.
We are using Orca as the screen reader.
I'm not familar with orca source code.
Im facing an issue and hope someone might be able to help me out.

We want orca to change the voice selected dynamically depending on the text that is available. Say, if we have opened an English text it should read it out in English using the voice kal_diphone, if the text is in some Indian languages, it should select the corresponding voice and read out.
For this, which files of orca should I be modifying ?
Can anybody give me an idea how to proceed with this.

Thanks a lot,
Anila.