Voice input helped me recover in the past. It was a challenge to get a workable setup, and getting back to the same speed takes a lot of tweaking and experimenting.
Sadly the CPU load and maintaining the setup with ever changing OSes has been too much. Though I do still maintain easier keyboard shortcuts almost religiously. AutoHotKey and Karabinder FTW.
I've designed Talon (my voice / alt input project) from the ground up for cross-platform scripts/configs. Once I port the OS layer to Windows/Linux you should be able to reuse mostly everything (with some app/os-specific stuff overridden at whatever abstraction layer it makes sense).
What sort of CPU load are you seeing? I'm able to dictate on battery, peaks averaging ~20% cpu. (Dragon is a different beast, peaking to >80% anytime it's over the noise threshold, so you need a really good mic to get good battery life).
I think that if you made a module that did voice commands for VLC media player that it could be quite popular. I've often wished I could just yell out "VLC pause" instead of having to walk over to the keyboard or mouse.
Sadly the CPU load and maintaining the setup with ever changing OSes has been too much. Though I do still maintain easier keyboard shortcuts almost religiously. AutoHotKey and Karabinder FTW.
Perhaps dedicated voice hardware can help.