@HarryVolek Sir, i have model for our PyTorch_Speaker_Verification ( epochs: 950 Max training speaker epoch for TIMIT)
sir i have a doubt. In that pytorch model, we will finetune(transfer learning/ Retrain) with my own speakers audios again.
Is it possible or not sir? which way to implement this?
AND another doubt, if we have only 10-20 speakers audio utterence, will it gives good accuracy or not?
(or)
in this 10-20 speakers audio utterence to build own model, will it gives good accuracy or not?
In that Voice Activity Detection how can we achieve "Wake word detection" sir? then how can i do this in dynamic way?
i am more concentrate on this. thank you for your initiative.
thank you lots @HarryVolek sir
@HarryVolek Sir, i have model for our PyTorch_Speaker_Verification ( epochs: 950 Max training speaker epoch for TIMIT) sir i have a doubt. In that pytorch model, we will finetune(transfer learning/ Retrain) with my own speakers audios again. Is it possible or not sir? which way to implement this?
AND another doubt, if we have only 10-20 speakers audio utterence, will it gives good accuracy or not? (or) in this 10-20 speakers audio utterence to build own model, will it gives good accuracy or not?
In that Voice Activity Detection how can we achieve "Wake word detection" sir? then how can i do this in dynamic way?
i am more concentrate on this. thank you for your initiative. thank you lots @HarryVolek sir