In this report we are going to address the issues of speech synthesis in changing noise conditions. We will investigate the potential improvements that can be introduced by using a speech synthesiser in these conditions that is able to modulate between a normal speech style and a speech style produced under noise. When we were faced with the selection of a research topic for this dissertation, the main initial idea was to find an interesting application related to reactive speech synthesis after being presented with a couple of reactive prototypes developed at the Centre for Speech Technology Research here in Edinburgh. The choice of reactivity to noise was motivated by the fertile bibliography on speech in noise and the latest technological improvements on parametric speech synthesis and intelligibility in noise. This report will contain a review of the related academic literature on this topic to contextualise our investigation and give a better understanding of our goals and the different methods that were used. Then we will present the main research goals of this investigation and two experiments that we devised to accomplish them: i) a production experiment of speech in changing noise conditions that focuses on the acoustic transformations that are generated to achieve the modulated speech styles, and ii) a listening evaluation with participants that compares the performance of an oracle speech synthesis system reactive to noise with two non-reactive systems. Finally, we will make a proposal for further work to implement a reactive system to noise, and some necessary precautions and potential features that could be interesting to include.