In many scenarios, people who share the same space need to listen to their own choice of audio with minimum interference from others. While headphones are a possible solution, they isolate people and make conversations difficult. Instead, we use loudspeakers to create sound zones for personalized audio delivery. Specifically, we apply audio filters and loudspeaker beamforming techniques to enhance the acoustic separation between sound zones. To enable head tracking for sound zones, we redirect the sound beams at high frequencies and leverage deep learning at low frequencies to adapt filters to new head positions in real time. We show in a demo with actual audio recordings that the proposed beamforming techniques can provide sufficient acoustic separation between sound zones.