Sam-hesta

Results 1 comments of Sam-hesta

> To bring more precision: actually 25s for a single audio file may be right as there is quite a big overhead for building the model (~20s). Once the model...