Good question, Robert.
Sorry but the answer is probably - it depends. It's not really a fair comparison, cause it's really two different methods. Both methods are very important for performance reasons and are often used in conjunction with each other.
The trigger-based approach needs manual setup and completely stops the processing of that sound regardless of the status of other sounds in your project. Virtual voice is more of a project-level prioritization system, that looks at all sounds and decided which ones are allowed to play, based on multiple priority parameters. For instance, if you are surrounded by 100s of enemies, you can use virtual voice to only play the 20 most relevant enemy sounds. This will save performance, but probably also make your soundscape less cluttered.
Maybe this blog post will clear up a few things too
https://blog.audiokinetic.com/en/how-to-get-a-hold-on-your-voices-optimizing-for-cpu-part-1/