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

[hts-users:03040] Re: How to use STC


Hi,
 
Then we could convert the full covariance to diagonal covariance?
But I suppose that would make there be no differences between the engine models...
So Anil, what is the advantage of STC? Theoretically, could we introduce this technique into diagonal covariance?

--
Na Xingyu (那兴宇)
Beijing Institute of Technology
naxy(at)bit.edu.cn
asr.naxingyu(at)gmail.com
naxingyu at {facebook, twitter, linkedin}


At 2011-09-23 15:54:13,"Keiichiro Oura" <uratec@xxxxxxxxxxxxxxx> wrote: >Hi, > >hts_engine file format for full covariance is not defined yet. >So, even if you untie the STC structure, STC voices for hts_engine >cannot be generated. > >Regards, >Keiichiro Oura > > >2011/9/23 那兴宇 <nxy-yzqs@xxxxxxx>: >> Hi Anil, >> >> I'm thinking that if you could untie the semi-tied structure of STC and >> re-clustere the model set so that it could be converted to engine format? >> That might affect the performance of STC but would be a slight promotion? >> I haven't checked the details of STC models so forgive me if I got it wrong. >> >> cheers, >> -- >> Na Xingyu (那兴宇) >> Beijing Institute of Technology >> naxy(at)bit.edu.cn >> asr.naxingyu(at)gmail.com >> naxingyu at {facebook, twitter, linkedin} >> >> >> At 2011-09-23 13:29:05,"Keiichiro Oura" <uratec@xxxxxxxxxxxxxxx> wrote: >>>Hi, >>> >>>hts_engine API does not support STC models. >>> >>>Regards, >>>Keiichiro Oura >>> >>> >>>2011/9/23 Anil John M <aniljohn80@xxxxxxxxx>: >>>> Hi All, >>>> >>>> This is my first question to the mailing list. My apologies if this is >>>> too naive. >>>> >>>> I have build an HTS voice for my data using the demo scripts of SLT. >>>> When I observed the synthesized files of alice sentences in the gen/ >>>> folder, I found STC based samples are sounding smooth than hts_engine >>>> samples. >>>> >>>> The models that are generated in voice/ directory can be used with >>>> hts_engine_api with respective switches for real-time synthesis. But I >>>> do not know where the STC models will be generated and how to use >>>> those models for real-time synthesis. I will be helpful if any one can >>>> provide this insight with a simple example. >>>> >>>> Thank you, >>>> - Anil >>>> >>>> >>> >> >> >> >



References
[hts-users:03039] Re: How to use STC, Keiichiro Oura
[hts-users:03036] How to use STC, Anil John M
[hts-users:03037] Re: How to use STC, Keiichiro Oura
[hts-users:03038] Re: How to use STC, 那兴宇