# Evaluating the Effectiveness of Transformer Layers in Wav2Vec 2.0, XLS-R, and Whisper for Speaker Identification Tasks

## TL;DR

This paper compares Wav2Vec 2.0, XLS-R, and Whisper speech models, analyzing their layer-wise features and optimal configurations for speaker identification, revealing how each model captures speaker-specific information at different depths.

## Contribution

It provides a detailed layer-wise analysis of three speech models for speaker identification and identifies optimal transformer layer configurations for each.

## Key findings

- Wav2Vec 2.0 and XLS-R capture speaker features in early layers.
- Fine-tuning enhances model stability and performance.
- Whisper performs better in deeper layers.

## Abstract

This study evaluates the performance of three advanced speech encoder models, Wav2Vec 2.0, XLS-R, and Whisper, in speaker identification tasks. By fine-tuning these models and analyzing their layer-wise representations using SVCCA, k-means clustering, and t-SNE visualizations, we found that Wav2Vec 2.0 and XLS-R capture speaker-specific features effectively in their early layers, with fine-tuning improving stability and performance. Whisper showed better performance in deeper layers. Additionally, we determined the optimal number of transformer layers for each model when fine-tuned for speaker identification tasks.

---
Source: https://tomesphere.com/paper/2509.00230