A Unit Selection-based Speech Synthesis Approach for Mandarin Chinese

The paper presents a unit selection-based speech synthesis approach for mandarin Chinese. Unit selection-based approach generates speech by selecting proper units from a speech corpus and connecting them together. In this approach, a set of features are defined to describe the speech units in the corpus and the expected units in the synthesized utterance. Based on the features, cost function is defined to select a sequence of units that are able to generate high quality speech. The cost function describes the two aspects of the generated speech, ie. (1) the appropriateness level of each unit itself. (2) the smoothness level between two units to be concatenated. Viterbi search algorithm is used in this approach to find the best unit sequence that minimizes the two costs. The authors use a new prosody description to ensure the prosody quality of the generated speech. Experiment shows that this approach can generate very high quality speech.