[hts-users:02661] Re: About monophone adaptation

[Simon King <Simon.King@xxxxxxxx>]

On 27 Dec 2010, at 01:40, 那兴宇 wrote:

> Do you mean that you want to convert the trained fullcontext HMMs to mono HMMs?
> If so, I think you can cluster the HMMs in mmf file according to the patterns of the models.

Yes - just untie everything, then re-tie. If you want to get monophones, you can do the re-tying without a decision tree, just write the HHEd "TI" commands yourself.

Simon


-- 
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.