Although the strings can be translated, the "Auto" setting means the right version probably won't be read.
It would be better to use the MathML and let the speech resolution and translated speech rules do the speech.
There are two places to change: relative rate change and relative pause change. A way to do this is to save the old preference value, set a new value, do the speech, and then restore the old value. A problem with this approach is that this preference is now (during the current instance of NVDA running) an API pref that overrides a user pref, so Apply/Ok needs to set the value because re-reading the pref file won't change the current value.
Although the strings can be translated, the "Auto" setting means the right version probably won't be read.
It would be better to use the MathML and let the speech resolution and translated speech rules do the speech.
There are two places to change: relative rate change and relative pause change. A way to do this is to save the old preference value, set a new value, do the speech, and then restore the old value. A problem with this approach is that this preference is now (during the current instance of NVDA running) an API pref that overrides a user pref, so Apply/Ok needs to set the value because re-reading the pref file won't change the current value.