hi!
what is the common way of splitting the corpus into a train and a test set? are there any conventions on that?
> what is the common way of splitting the corpus into a train and a test set?
In Voxforge sets each prompt from 10 in the full list go to the test
> are there any conventions on that?
There is no common convention like that. In other corpuses the test set is created differently.
sorry, could you please be more precise on "each prompt from 10 in the full list"?