[hts-users:00988] Re: Binary file size of HTS

["Heiga ZEN (Byung Ha CHUN)" <zen@xxxxxxxxxxxxxxx>]

Hi,

Sacha Krstulovic wrote (2007/11/21 1:07):

> To add to Sungho's question:
> from the size of the model stored on disk, how much should be
> roughly added or multiplied to obtain the final synthesis engine
> footprint, which accounts for the allocation of internal
> data structures?

It's hard to tell.
To obtain a waveform, first we load models and trees, and then run the speech parameter generation algorithm and the MLSA filter.

Memory footprint to store models are almost the same as those of *.pdf files.
Memory footprint to store trees are smaller than *.inf files because *.inf files are in ASCII but most of them are loaded in binary.
Memory footprint required by the speech parameter generation algorithm depends on the utterance length and vector size.
Memory footprint required in the MLSA filter is very small compared to the other parts.

Regards,

Heiga ZEN (Byung Ha CHUN)

--
------------------------------------------------
Heiga ZEN     (in Japanese pronunciation)
Byung Ha CHUN (in Korean pronunciation)

Department of Computer Science and Engineering
Nagoya Institute of Technology
Gokiso-cho, Showa-ku, Nagoya 466-8555 Japan

http://www.sp.nitech.ac.jp/~zen
------------------------------------------------