Person name is intrinsically difficult to pronounce due to its variety of origin and non-regularity. This poses significant challenges to contact dialing by voice. In order to mitigate the pronunciation problem for contact names, we propose using personalized pronunciation learning: people can use their own pronunciations for their contact names. We achieve this by implicitly learning from users' corrections in real time, providing a seamless user experience improvement. We show that personalized pronunciation significantly reduces word error for difficult contact names by 15% relatively.