Voice Quality Modification Using a Harmonics Plus Noise Model: Transferring Vocal Effort with Parallel Corpora

The harmonics plus noise model (HNM) has been used for prosodic speech signal modifications in high-quality environments in recent decades. Such speech modification techniques allow Text-To-Speech systems to generate more expressive synthesis without requiring extensive corpora resources. A more expressive synthesis can improve the user experience with Human-Machine-Interfaces. In this paper, an adaptation of the adaptive pre-emphasis linear prediction technique to the HNM for modifying vocal effort is presented. The proposed transformation methodology is validated using a copy re-synthesis strategy on a speech corpora specifically designed for vocal effort research. The perceptual tests demonstrate the effectiveness of the proposed technique in performing various types of vocal effort conversions for the given corpus.