Speaker Verification with Speech-Aware LLMs: Evaluation and Augmentation

TL;DR

This paper evaluates speech-aware large language models for speaker verification, revealing their limited speaker discrimination ability and proposing an augmentation method that significantly improves verification performance while maintaining language capabilities.

Contribution

It introduces a model-agnostic scoring protocol for speaker verification and a lightweight augmentation method to enhance LLMs with speaker verification capabilities.

Findings

Weak speaker discrimination in current speech-aware LLMs (EER > 20%)

Proposed augmentation improves EER to 1.03% on VoxCeleb1-E

Method preserves natural language interface while adding speaker verification

Abstract

Speech-aware large language models (LLMs) can accept speech inputs, yet their training objectives largely emphasize linguistic content or specific fields such as emotions or the speaker's gender, leaving it unclear whether they encode speaker identity. First, we propose a model-agnostic scoring protocol that produces continuous verification scores for both API-only and open-weight models, using confidence scores or log-likelihood ratios from the Yes/No token probabilities. Using this protocol, we benchmark recent speech-aware LLMs and observe weak speaker discrimination (EERs above 20% on VoxCeleb1). Second, we introduce a lightweight augmentation that equips an LLM with ASV capability by injecting frozen ECAPA-TDNN speaker embeddings through a learned projection and training only LoRA adapters. On TinyLLaMA-1.1B, the resulting ECAPA-LLM achieves 1.03% EER on VoxCeleb1-E, approaching a dedicated speaker verification system while preserving a natural-language interface.