Hi. I have been trying to create a model for single digit detection, but I kept it at word-level (no phones). What I wanted to do was incorporate the silence model.

I have done everything concerning the silence HMM. When I use the "sil" label in my transcription, the silence is not recognised and it is deleted (all instances passed to the "deletions") column, which basically ruins the performance of the model. When I remove all the labels from the transcription, I get 100% recognition. Can anyone help with setting up the model so that I can recognise the silences?

Thank you!!