I want to build my own TTS (Text to Speech) App using HTS (HMM-based speech synthesis system) for the Arabic language. I fail to find any step by step instructions on how to build the synthesizer using HTS. What I have done is to download the sample Speaker Dependent Demo on the HTS website and train that data ana tested it on Festival (English speaker). Now I don't know what files should I change in the HTS-demo to build my voice(language).
You first build Festival unit selection voice for your language following Building Synthetic Voices guide.
After you have the voice and required lab and utt files, you run training.pl script from HTS with updated paths to your database and it will build the voice for you.