The hts_engine_API is a fast synthesizer allowing real-time TTS using HTS model. Through parsing the model structure it uses, i.e. ".pdf" file, new researchers in this field can quickly and briefly understand the key technique. This parser is to visualize the binary model file to demonstrate the interval structure.
- I choose to parse the model used by API version 1.03/4/5/6, which was generated by HTS-demo scripts version 2.1.1 and 2.2, because it contains most information.
- From API version 1.03, the model header information includes MSD flag, stream number and vector length (static+dynamic).
- From API version 1.07, header information is removed from model file, and obtained by parsing a model information file.