[Subject Prev][Subject Next][Thread Prev][Thread Next][Date Index][Thread Index]

[hts-users:01826] Re: HTS voice built with Festvox 2.1 - HHed failure during duration model clustering


Hallo,

thank you for reply. However, I have compiled all the tools on my own (on 
linux x86-64, or former amd64). I suppose, therefore, that correct endian has 
automatically been chosen based on the platform. And all the files mentioned 
were generated by the compiled tools.

Moreover, I gues that I would need to know exact format of the .pdf files to 
convert them (if it is anything more complicated than a sequence of float 
numbers ...).

Are you sure that the conversion is the key?

Thank you,
Dan


On Friday 12 of December 2008, Keiichiro Oura wrote:
> Hi, Daniel
>
> > Now the question is, how to "map" the files generated into Festival
> > voice. I tried the following mapping, but without success - the voice was
> > loaded correctly, but SayText("Hallo, this is the first try") command
> > consumed about 1.5GB memory and failed with message WALLOC: failed to
> > malloc 671088648 bytes.
>
> I think that it's an endian problem.
>
> software   : file format
> -----------:--------------
> HTK        : big endian
> HTS        : big endian
> hts_engine : big endian
> festival   : little endian
>
> > - duration.pdf		--> voices/qst001/ver1/dur.pdf
> >   simply renamed
>
> You should 'swap' command in SPTK to change endian.
>
> example)
> % swap +f duration.pdf > voices/qst001/ver1/dur.pdf
>
> Regards,
> Keiichiro Oura


Attachment: signature.asc
Description: This is a digitally signed message part.


Follow-Ups
[hts-users:01827] Re: HTS voice built with Festvox 2.1 - HHed failure during duration model clustering, Heiga ZEN (Byung Ha CHUN)
References
[hts-users:01817] HTS voice built with Festvox 2.1 - HHed failure during duration model clustering, Daniel Tihelka
[hts-users:01824] Re: HTS voice built with Festvox 2.1 - HHed failure during duration model clustering, Daniel Tihelka
[hts-users:01825] Re: HTS voice built with Festvox 2.1 - HHed failure during duration model clustering, Keiichiro Oura