training a base model

[ppisljar]

I am trying to use this in a new languge.
I have robert model, i wrote g2p function and i have 500- hours of audio with transcriptions

#245 (comment)

here i read that i should turn of DD and SLM D .... duration descriminator is easy, there is a flag in config file to turn it off.
but how can i turn off slm discriminator ?

thank you for your help/