In this paper, potential synthesis techniques for mammalian vocalisation sound effects are analysed. Physically-inspired synthesis models are devised based on human speech synthesis techniques and research into the biology of a mammalian vocal system. The benefits and challenges of physically-inspired synthesis models are assessed alongside a signal-based alternative which recreates the perceptual aspects of the signal through subtractive synthesis. Nonlinear aspects of mammalian vocalisation are recreated using frequency modulation techniques, and Linear Prediction is used to map mammalian vocal tract configurations to waveguide filter coefficients. It is shown through the use of subjective listening tests that such models can be effective in reproducing harsh, spectrally dense sounds such as a lion’s roar, and can result in life-like articulation.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.