# Evaluating Bias in Spoken Dialogue LLMs for Real-World Decisions and Recommendations

arXiv: https://arxiv.org/abs/2510.02352 · HTML: https://mlsanity.com/paper/2510.02352/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Builds on 32 of its 36 references that are in the map.
- First posted on arXiv 27 Sep 2025.

## Builds on (most cited first, 30 of 32)

- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [MultiWOZ -- A Large-Scale Multi-Domain Wizard-of-Oz Dataset for Task-Oriented Dialogue Modelling](https://mlsanity.com/paper/1810.00278/) (2018)
- [CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models](https://mlsanity.com/paper/2010.00133/) (2020)
- [Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods](https://mlsanity.com/paper/1804.06876/) (2018)
- [DailyDialog: A Manually Labelled Multi-turn Dialogue Dataset](https://mlsanity.com/paper/1710.03957/) (2017)
- [Bias and Fairness in Large Language Models: A Survey](https://mlsanity.com/paper/2309.00770/) (2023)
- [Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models](https://mlsanity.com/paper/2311.07919/) (2023)
- [Moshi: a speech-text foundation model for real-time dialogue](https://mlsanity.com/paper/2410.00037/) (2024)
- [GLM-4-Voice: Towards Intelligent and Human-Like End-to-End Spoken Chatbot](https://mlsanity.com/paper/2412.02612/) (2024)
- [Gender bias and stereotypes in Large Language Models](https://mlsanity.com/paper/2308.14921/) (2023)
- [LLaMA-Omni: Seamless Speech Interaction with Large Language Models](https://mlsanity.com/paper/2409.06666/) (2024)
- [Is ChatGPT Fair for Recommendation? Evaluating Fairness in Large Language Model Recommendation](https://mlsanity.com/paper/2305.07609/) (2023)
- [Data-efficient Fine-tuning for LLM-based Recommendation](https://mlsanity.com/paper/2401.17197/) (2024)
- [Bias and Unfairness in Information Retrieval Systems: New Challenges in the LLM Era](https://mlsanity.com/paper/2404.11457/) (2024)
- [IndexTTS2: A Breakthrough in Emotionally Expressive and Duration-Controlled Auto-Regressive Zero-Shot Text-to-Speech](https://mlsanity.com/paper/2506.21619/) (2025)
- [EARS: An Anechoic Fullband Speech Dataset Benchmarked for Speech Enhancement and Dereverberation](https://mlsanity.com/paper/2406.06185/) (2024)
- [SLAM-Omni: Timbre-Controllable Voice Interaction System with Single-Stage Training](https://mlsanity.com/paper/2412.15649/) (2024)
- [LLM Voting: Human Choices and AI Collective Decision Making](https://mlsanity.com/paper/2402.01766/) (2024)
- [FairMT-Bench: Benchmarking Fairness for Multi-turn Dialogue in Conversational LLMs](https://mlsanity.com/paper/2410.19317/) (2024)
- [Toward Pareto Efficient Fairness-Utility Trade-off inRecommendation through Reinforcement Learning](https://mlsanity.com/paper/2201.00140/) (2022)
- [Towards measuring fairness in speech recognition: Fair-Speech dataset](https://mlsanity.com/paper/2408.12734/) (2024)
- [Hey ASR System! Why Aren't You More Inclusive? Automatic Speech Recognition Systems' Bias and Proposed Bias Mitigation Techniques. A Literature Review](https://mlsanity.com/paper/2211.09511/) (2022)
- [Spoken Stereoset: On Evaluating Social Bias Toward Speaker in Speech Large Language Models](https://mlsanity.com/paper/2408.07665/) (2024)
- [An Audit of Misinformation Filter Bubbles on YouTube: Bubble Bursting and Recent Behavior Changes](https://mlsanity.com/paper/2203.13769/) (2022)
- [MuTox: Universal MUltilingual Audio-based TOXicity Dataset and Zero-shot Detector](https://mlsanity.com/paper/2401.05060/) (2024)
- [MMDT: Decoding the Trustworthiness and Safety of Multimodal Foundation Models](https://mlsanity.com/paper/2503.14827/) (2025)
- [A Multi-Stage Multi-Codebook VQ-VAE Approach to High-Performance Neural TTS](https://mlsanity.com/paper/2209.10887/) (2022)

## Built on by (most cited first, 7 of 7)

- [VoxSafeBench: Not Just What Is Said, but Who, How, and Where](https://mlsanity.com/paper/2604.14548/) (2026)
- [Counterfactual Fairness Evaluation of LLM-Based Contact Center Agent Quality Assurance System](https://mlsanity.com/paper/2602.14970/) (2026)
- [From Seeing it to Experiencing it: Interactive Evaluation of Intersectional Voice Bias in Human-AI Speech Interaction](https://mlsanity.com/paper/2604.13067/) (2026)
- [Voice or Stereotype? Disentangling Acoustic and Content-Based Gender in Speech-to-Speech Models](https://mlsanity.com/paper/2609.09263/) (2026)
- [Measuring Fairness in Large Audio Language Models via Semantic-Aware Bias Estimation](https://mlsanity.com/paper/2608.13624/) (2026)
- [Auditing Bias and Safety in Voice AI Customer Care](https://mlsanity.com/paper/2609.04206/) (2026)
- [Personalization Without Stereotyping: Do Speech LLMs Know When to Turn a Deaf Ear?](https://mlsanity.com/paper/2610.13550/) (2026)

## Abstract (arXiv)

While biases in large language models (LLMs), such as stereotypes and cultural tendencies in outputs, have been examined and identified, their presence and characteristics in spoken dialogue models (SDMs) with audio input and output remain largely unexplored. Paralinguistic features, such as age, gender, and accent, can affect model outputs; when compounded by multi-turn conversations, these effects may exacerbate biases, with potential implications for fairness in decision-making and recommendation tasks. In this paper, we systematically evaluate biases in speech LLMs and study the impact of multi-turn dialogues with repeated negative feedback. Bias is measured using Group Unfairness Score (GUS) for decisions and similarity-based normalized statistics rate (SNSR) for recommendations, across both open-source models like Qwen2.5-Omni and GLM-4-Voice, as well as closed-source APIs such as GPT-4o Audio and Gemini-2.5-Flash. Our analysis reveals that closed-source models generally exhibit lower bias, while open-source models are more sensitive to age and gender, and recommendation tasks tend to amplify cross-group disparities. We found that biased decisions may persist in multi-turn conversations. This work provides the first systematic study of biases in end-to-end spoken dialogue models, offering insights towards fair and reliable audio-based interactive systems. To facilitate further research, we release the FairDialogue dataset and evaluation code.