First of all, I'm not a developer of the HTS system.
So, I'm not sure what I'm saying is right or not.
question 1
Is "hts_engine" better than the SPTK (in the "HTS-demo" , it uses SPTK )
when systhsizing the speech?
The c codes of "SPTK", "HTS" and "hts_engine" are opened in public.
If you look into them and compare them, you will find the answer.
They are basically same.
question 2
Are the .inf files and .pdf files more efficient than HMM files?
If you check the perl code, "
Training.in" file, you will find the answer.
xxx.pdf files are HMM files.
More efficient?
I'm not the developer, but at least I can say, binary format is faster then ascii format to load.
Good luck.