Speech synthesis systems are concerned with generating a natural sounding and intelligiblernSpeech by taking text as input. Speech Synthesizers are very important in helping impairedrnpeople, in teaching and learning process, for telecommunications and industries. Though itrnhas many applications, generating intelligible and natural sounding synthetic speech has beenrna challenging task for years. To overcome these challenges, different techniques have beenrnstudied and implemented.rnThough speech synthesizers based on HMM are done for foreign languages, they are notrnapplicable for Amharic language since the languages special characteristics are notrnconsidered in these synthesizers. Hence, in this thesis work Hidden Markov Model basedrnspeech synthesis for Amharic language (HTS-FA) is done.rnThe HTS-FA has two phases: the training and synthesis phase. The main activities included inrnthe training phase are preparation of the training dataset, language modeling, featurernextraction and training the model. In the synthesis phase, models are selected according to therntext to be synthesized, and then speech parameters are generated from them. Finally, thernsynthesized speech is generated from the speech parameters.rnA total of five hundred sentences are used for training the model from a corpus having a sizernof 11,670 sentences, and twenty sentences, which are not included in the training dataset, arernused for testing the performance of the system. In this thesis, the Mean Opinion Score (MOS)rnevaluation technique is used. The results from the MOS were found to be 4.12 and 3.6 forrnintelligibility and naturalness respectively for speeches synthesized by HTS-FA. Usingrnxiirnconcatinative method the result obtained for intelligibility and naturalness are 3.54 and 3.25rnrespectively.rnKeywords: Speech synthesis, HMM, HMM based speech synthesis, Language Modeling