Personalization Strategies for End-to-End Speech Recognition Systems