Bálint Tóth, Géza Németh
Hidden-Markov-Model based speech synthesis in Hungarian
This paper describes the first application of the Hidden-Markov-Model based text-to-speech synthesis method to the Hungarian language. The HMM synthesis has numerous favorable features: it can produce high quality speech from a small database, theoretically it allows us to change characteristics and style of the speaker and emotion expression may be trained with the system as well.