The formant speech synthesis methodology follows the simplified source-filter model. The rules to control the model are typically created by linguists to closely replicate the evolution of the formant structure and other spectral features of normal speech. Formant synthesis can produce clear speech with modest computational resources, making it suitable for embedded systems that don't depend on a large human speech corpus, unlike [[concatenative synthesis]]. Nonetheless, synthesized speech lacks naturalness and exhibits artifacts, while the development of rules for the synthesis process is challenging. ![[klatt-cascade-parallel.png]] Formant filter: ![[formant-frequency-and-bandwidth.png]] F - Formant frequency B - Formant bandwidth ![[klatt-synthesizer.png]] [Klatt 1980](http://sail.usc.edu/~lgoldste/Ling582old/Week%2012/klatt1980.pdf) ## References P Seeviour, J Holmes, and M Judd. Automatic generation of control signals for a parallel formant speech synthesizer. In ICASSP’76. IEEE International Conference on Acoustics, Speech, and Signal Processing, volume 1, pages 690–693. IEEE, 1976. Jonathan Allen, Sharon Hunnicutt, Rolf Carlson, and Bjorn Granstrom. Mitalk-79: The 1979 mit text-to-speech system. The Journal of the Acoustical Society of America, 65(S1): S130–S130, 1979. Dennis H Klatt. Software for a cascade/parallel formant synthesizer. the Journal of the Acoustical Society of America, 67(3):971–995, 1980.