Do I need to keep the 1G Natural Language component in order to use text-to-speech?

Could I just keep Speech Support?