[hts-users:03612] Re: Engine refresh issue

[Ekrem Guner <ekremgunerozu@xxxxxxxxx>]

One simple solution is to create one label for all sentences that you want to syntesize together (just concatenate them in one label file).

Another one (if you want more control) needs some efforts, changing the code. 

Engine first loads the voice model then loads the label and syntehsizes the speech for that label.

You can separate them as voice loading and synthesizing label as two functions.

You can call second function in a loop for each label. 

But be carefull about the memory leaks. 

You need to refresh the engine (de-allocated the necessary memory parts) at the end of the function before loading the new label. 

   

On Fri, Feb 15, 2013 at 11:39 AM, pramod kumar <pramod2001.india@xxxxxxxxx> wrote:

Hi,

I need to synthesize multiple sentences one by one. 

Currently, hts_engine is loading the models for each sentence. Is there a way to avoid the model loading for each sentence.

--

Thanks & Regards,

Pramod.