Details
Poster
Presenter(s)
![Xuechen Liu Headshot](https://confcats-catavault.s3.amazonaws.com/CATAVault/ieeecass/master/files/styles/cc_user_photo/s3/user-pictures/24161.png?h=0415e1b2&itok=DvuN7YDL)
Display Name
Xuechen Liu
- Affiliation
-
AffiliationUniversity of Eastern Finland / Université de Lorraine, CNRS, Inria
- Country
Abstract
We propose a learnable mel-frequency cepstral coefficient (MFCCs) frontend architecture for neural network based automatic speaker verification (ASV). Our architecture not only retains the simplicity and interpretability of MFCC-based features while allowing the model to be adapted to data flexibly, but also leads to performance improvement on ASV system performance.