[hts-users:00890] Re: Training by imposing decision trees

[Alexis Moinet <alexis.moinet@xxxxxxxxxx>]

Simon King wrote :
> Heiga ZEN (Byung Ha CHUN) wrote:
> > In this case parameters of generated MMF are equal to those of MMF 
> > used to construct decision trees.
> > So you should re-estimate model parameters 
>
> Good point - I hadn't considered the case where the decision tree 
> being imposed might have been trained on different data, I just 
> thought about the case where it matched the data - although the case 
> of differing data is obviously more likely and more interesting.

In our case, that's exactly what we want to do.

To be exact, we simply want to have the same trees for two different 
speakers at the end of the two trainings and we thought that training 
one speaker and then imposing its tree to the other speaker's training 
would be the trivial solution to achieve this.

Now it doesn't seem that simple anymore :-)

We will give it a try with your solutions an keep you informed of the 
results,

thank you very much,

Alexis