Nonlinear emotional prosody generation and annotation

摘要: Emotion is an important element in expressive speech synthesis. The paper makes the brief analysis on prosody parameters, stresses, rhythms and paralinguistic information different emotional speech, labels with rich annotation multi-layers. Then, a CART model used to do generation. Unlike traditional linear modification method, which direct of F0 contours syllabic durations from acoustic distributions such as, topline, baseline, intensities, models try map subtle between neutral within various context information. Experiments show that, model, able generate good outputs, however results could be improved if more information, as breaks jitter are integrated into