A system and application for real-time preferred speaker identification using a mobile device in combination with a hearing system, based on a fine-tuned speech model, customized voice print / voice biometrics and source separation for multi-speaker and noisy environments.