Available model is much larger than model discussed in the paper

#11
by mcorb1 - opened

Hi there!

I have a question regarding the size of the model. In the paper: ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation
in TDNN Based Speaker Verification. The discussed model is:
ECAPA-TDNN (C=512) 6.2M
ECAPA-TDNN (C=1024) 14.7M

However when I load the model it has 22.2M parameters. Could anyone explain this?

Look forward to hearing from you!
Marcus

Sign up or log in to comment