Although R&D into ‘speech synthesis’ has received a considerable amount of attention over many years, there has been remarkably little effort devoted to constructing vocal synthesisers for non-human animals. Of course, interest in synthesising human speech has been driven by the demand for practical applications such as reading machines for the blind or voice-operated assistants. Nevertheless, there are potential uses for non-human vocal synthesis: e.g. in education, robotics or ecological fieldwork. The latter is of particular interest, since it is common practice to use ‘playback’ methods (based on recorded samples) that do not easily facilitate parametric control over key experimental variables. Therefore, this paper presents the design and implementation of a real-time parametric general-purpose mammalian vocal synthesiser. The approach taken has been to decompose the overall sound production system into the relevant anatomical components (such as the lungs, vocal folds, tongue and mouth), and to implement a real-time simulation in ‘Pure Data’ — an open-source dataflow programming language. The software was successfully used to design an appropriate mammalian voice for the MiRo biomimetic robot, but there are potential applications in a number of areas. The software is available for free download at http://www.dcs.shef.ac.uk/~roger/downloads.html.
Cite as: Moore, R.K. (2016) A Real-Time Parametric General-Purpose Mammalian Vocal Synthesiser. Proc. Interspeech 2016, 2636-2640, doi: 10.21437/Interspeech.2016-841
@inproceedings{moore16_interspeech,
author={Roger K. Moore},
title={{A Real-Time Parametric General-Purpose Mammalian Vocal Synthesiser}},
year=2016,
booktitle={Proc. Interspeech 2016},
pages={2636--2640},
doi={10.21437/Interspeech.2016-841}
}