- Author: Emil-18
- NVDA compatibility: 2026.1
- Download: Stable version.
This add-on Automaticly detects the language of the text NVDA is about to speak, and uses NVDA's built in auto language switching, if turned on, to sspeak the text in that language.
This add-on is heavily inspired by the language ident add-on. That add-on stopped working, however, and from it's issues on GitHub, it doesn't seam like the developer is going to fix it.
- language detection library:
This is a combo box that allows you to change the library the add-on uses for language detection. The choices are:
- lingua (recommended)
- Langdetect
- Language interpretation:
This is a combo box that allows you to select when the add-on is interpreting the language of the text. The choices are:
- always interpret: the add-on will always interpret the language of the text and remove all previous instructions to change the language. For example, if you are on a web page, and the language of the web page is defined as english, but according to the add-on, the text is Norwegian, it will switch to a Norwegian voice. If you haven't selected Norwegian as one of the languages to be interpreted, (see below), the voice will not change at all.
- only interpret if the text doesn't instruct NVDA to change language: The add-on will only interpret the language if there are no instructions to change the language in the text. In the example above, it will switch to an english voice.
- never interpret (NVDA's default behavior): Self explanatory.
- Detect multiple languages in the same text: this is a check box that allows you to specify if the add-on should try to detect multiple languages in the same text chunk that NVDA is about to speak. For example, if you are reading a document line by line, and encounter a line containing two languages, it will try to detect each language Separately.
- Number of words required to switch language. This is a numeric edit field that allows you to choose the minimum word count required to make the language detection activate. If the spoken text contains less words than this value, the add-on will behave as if it is disabled
- Languages to interpret: This is a list of the languages to interpret. If the add-on interprets a text as a language that isn't selected in this list, no auto language switching is done based on the interpretation. No languages are selected by default.
This version was made by @heath-toby.
- Added compatibility with NVDA 2026.1.
- Added a setting that allows you to specify the minimum amount of words required for the add-on to activate.
- Fixed a bug that caused speech to stop or NVDA to play error sounds when no language is selected
- Fixed a bug that caused all languages supported by lingua to be checked by default
- The russian translation has been updated, thanks to Kostenkov-2021
- Added support for the lingua language detection library. This library is much faster and much more correct than langdetect, wich was used previously. In addition, it will always interpret the same text in the same language, in contrast to langdetect. You will still be able to use langdetect if you need to for some reason, but this is not recommended
- Added compatibillity with NVDA 2025.
- Fixed a bug that caused the add-on to play error sounds when exiting NVDA.
- Made it so no language interpretation is performed (as it is verry slo) if the user has auto language switching turned off.
- Russian translation has been added, thanks to Kostenkov-2021
Initial release