# A Comparative Re-Assessment of Feature Extractors for Deep Speaker Embeddings

arXiv: https://arxiv.org/abs/2007.15283 · HTML: https://mlsanity.com/paper/2007.15283/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 30 Jul 2020.

## Built on by (most cited first, 5 of 5)

- [Tiny, always-on and fragile: Bias propagation through design choices in on-device machine learning workflows](https://mlsanity.com/paper/2201.07677/) (2022)
- [Learnable MFCCs for Speaker Verification](https://mlsanity.com/paper/2102.10322/) (2021)
- [Optimized Power Normalized Cepstral Coefficients towards Robust Deep Speaker Verification](https://mlsanity.com/paper/2109.12058/) (2021)
- [Parameterized Channel Normalization for Far-field Deep Speaker Verification](https://mlsanity.com/paper/2109.12056/) (2021)
- [Optimizing Multi-Taper Features for Deep Speaker Verification](https://mlsanity.com/paper/2110.10983/) (2021)

## Abstract (arXiv)

Modern automatic speaker verification relies largely on deep neural networks (DNNs) trained on mel-frequency cepstral coefficient (MFCC) features. While there are alternative feature extraction methods based on phase, prosody and long-term temporal operations, they have not been extensively studied with DNN-based methods. We aim to fill this gap by providing extensive re-assessment of 14 feature extractors on VoxCeleb and SITW datasets. Our findings reveal that features equipped with techniques such as spectral centroids, group delay function, and integrated noise suppression provide promising alternatives to MFCCs for deep speaker embeddings extraction. Experimental results demonstrate up to 16.3% (VoxCeleb) and 25.1% (SITW) relative decrease in equal error rate (EER) to the baseline.