The documentation contains major steps for building a HTS voice (version 2.2), namely, Hybrid Segmentation for segmenting speech into accurate phonetic units, Pruning for improving the quality of segmentation, HTS voice building and HTS with STRAIGHT vocoder voice building steps. The TBT toolkit covers all steps required to build TTS from start to end. It also provides the synthesis framework. HTS 2.3 Synthesis Using a Built Voice Model details how to synthesize Indian language text using an already built HTS 2.3 voice model.
Downloads
To download the source code, you must agree to the License For Use of IndicTTS. Click Agree to continue.
Indic TTS : Enhancing text-to-speech (TTS) synthesis for Indian languages, optimizing quality and integrating compact TTS into disability aids and diverse applications.