Abstract
Harmonics plus Noise Model (HNM) is a well known speech signal representation technique that allows to apply high quality modifications to the signal used in text-to-speech systems providing higher flexibility than its counterpart TD-PSOLA based synthesis systems. In this paper an adaptation of the adaptive pre-emphasis linear prediction technique for modifying the vocal effort, using HNM speech representation, is presented. The proposed transformation methodology is validated using a Copy Re-synthesis strategy on a speech corpora specifically designed with three levels of vocal effort (soft, modal and loud). The results of a perceptual test demonstrate the effectiveness of the proposed technique performing all different vocal effort conversions for the given corpus.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call