Ability for batched tensor

[MordehayM]

Thank you for the code!
I have a question: Does this code has the ability to calc the stoi for batched tensor with size of [B, num_of_samples] or even with size of [B, num_speaker, num_of_samples].
I checked and I think it has not this ability, right?
Can you maybe expand the implementaion to this scenario please?

[mpariente]

Thanks for the question !
Indeed, it's not currently possible.

There is a simple VAD within the STOI computation. All the steps afterwards cannot be batched because the number of samples of active segments within batch examples are not the same.

There is an attempt to do that here : torch_stoi, but batching the real computation does not work.
This note might be interesting to understand what I mean.

[mpariente]

I'm closing the issue as we won't implement any batched computation.