Really like the use of the voice sample and just the breakdown part in general around 3 minutes. I think it could have even been dragged out longer and / or had some panning and echo parts included to amp up the move to the next section.