Voice messages, by nature, prevent users from gauging the emotional tone without fully diving into the audio content. This hinders the shared emotional experience at the pre-retrieval stage. Research scarcely explored "Emotional Teasers"-pre-retrieval cues offering a glimpse into an awaiting message's emotional tone without disclosing its content. We introduce EmoWear, a smartwatch voice messaging system enabling users to apply 30 animation teasers on message bubbles to reflect emotions. EmoWear eases senders' choice by prioritizing emotions based on semantic and acoustic processing. EmoWear was evaluated in comparison with a mirroring system using color-coded message bubbles as emotional cues (N=24). Results showed EmoWear significantly enhanced emotional communication experience in both receiving and sending messages. The animated teasers were considered intuitive and valued for diverse expressions. Desirable interaction qualities and practical implications are distilled for future design. We thereby contribute both a novel system and empirical knowledge concerning emotional teasers for voice messaging.