Comment 1 by jteh on 2013-10-07 22:21
This is mostly covered by #667 except for two things:

1. NVDA uses the interface language instead of the speech language if automatic language switching is disabled. This is covered by Optionally use symbols/character descriptions for the speech language even when automatic language switching is disabled #3545.
2. You can't switch profiles using the synth settings ring, as the synth settings ring can't currently be used for anything other than synth settings. However, as part of Configuration profiles #667, you can easily switch profiles via the Configuration Profiles dialog (NVDA+control+p). Is this sufficient to satisfy your request?

Comment 2 by jorgtum on 2013-10-08 08:47
The shortcut is not important, is equals do nvda+ ctrl + arrow key or nvda + ctrl + p, this is indifferent.

Is not all covered, because the punctuation symbols must be readed in the language of the speech instead the interface language, but too the announcement of links, headers, images, lists, tables, columns etc should be announced in the speech language, when I am reading a English text, and I have the English voice, I need all be readed in English, for instance, if I have a link in the middle of a text, this must be readed like "lin" (English) and not like "vinclo" (Aragonese). The same case to all the language, if I'm reading a French text, I need that all be pronnounced in French, etc.

Thanks and sorry by repeat.

Jorge.

Comment 3 by leonarddr on 2013-10-08 08:52
This is because the words for link, table etc. are part of the NVDA interface language. Interface language could only be changed by setting the language option in general settings and restarting NVDA.
I don't know why this restarting is required, probably because language stuff couldn't be re-initialized without a restart. Although it would be a cool future to switch interface language on the fly, this probably isn't possible.

Comment 4 by jteh on 2013-10-15 07:01
Switching interface language without a restart isn't currently possible because the strings are translated at the time they are processed and many of them are processed when the code is loaded. Even if it were made possible, using the content language for control and formatting information but the interface language for the rest of the UI is even harder, as these strings are translated in the same way. If that's what you are requesting (and the other issues are covered elsewhere), we can morph this ticket accordingly, but it's worth noting that this would be extremely low priority and unlikely to be implemented by us. Still, someone else might want to do it.

Comment 5 by jorgtum on 2013-10-15 08:10
I don't need change all the interface language, I only need when the voice changes to another language, the messages that reports about the navigation be used on this language, how I saw, the report of links, headers of level N, the punctuation symbols, the report of lists, graphics, tables, etc. else how it is now say that messages on the language that be the interface and with the pronnounce of the current voice.

I think the priority is that the punctuation symbols (colon, dot, semicolon ...) are readed on the same language,being in a different file it should be easier, this must be changed in fly when the voice changes, the announce of other messages like links, headers etc have less priority but I think also is important.

I think this could be acomplished with speech refactor, if you set the same synth as second synth to pronounce only the element types and status. Then you must link the first synth with the second one and if you change the voice language for synth 1 the language for synth 2 changes too. Is this possible from a programatical point of view? The most complicated thing is actually to manages the interactions between the voices. Synth 1 should make a pause of x ms and in this pause synth 2 should report the element type. After that, synth 1 should continue reading the displayed text until next element type or status is identified.
I don't know if this is possible for all applications but this would make sense also in MS Excel for example.

I think also a problem is to separate language of reporting of element type and status from the interface language.

Maybe speech refactoring is not required at all if element type and status is separated from interface language.