Catalogue of Artificial Intelligence Techniques
Author(s): Steve Isard
Basic sounds of a language can be characterised by their Formant frequencies. The formant pattern for a synthetic utterance can be created by interpolating between the formant values of the sounds composing the desired utterance. A speech wave can be computed from such a formant pattern, to be played through a digital-to-analogue converter, or the formant pattern can be used directly to excite a set of resonators, acting as an electrical analogue of the vocal tract.
- Klatt, D.H., Software for a cascade/parallel formant synthesizer Journal of Acoustic Society of America 67 (1980), 971--995.