# Bias in the Ear of the Listener: Assessing Sensitivity in Audio Language Models Across Linguistic, Demographic, and Positional Variations

arXiv: https://arxiv.org/abs/2602.01030 · HTML: https://mlsanity.com/paper/2602.01030/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 37 references that are in the map.
- First posted on arXiv 1 Feb 2026.

## Builds on (most cited first, 17 of 17)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://mlsanity.com/paper/2111.09296/) (2021)
- [SpeechGPT: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities](https://mlsanity.com/paper/2305.11000/) (2023)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [VoiceBench: Benchmarking LLM-Based Voice Assistants](https://mlsanity.com/paper/2410.17196/) (2024)
- [Having Beer after Prayer? Measuring Cultural Bias in Large Language Models](https://mlsanity.com/paper/2305.14456/) (2023)
- [Spoken Question Answering and Speech Continuation Using Spectrogram-Powered LLM](https://mlsanity.com/paper/2305.15255/) (2023)
- [Dialect prejudice predicts AI decisions about people's character, employability, and criminality](https://mlsanity.com/paper/2403.00742/) (2024)
- [Twists, Humps, and Pebbles: Multilingual Speech Recognition Models Exhibit Gender Performance Gaps](https://mlsanity.com/paper/2402.17954/) (2024)
- [SSR: Alignment-Aware Modality Connector for Speech Language Models](https://mlsanity.com/paper/2410.00168/) (2024)
- [Assessment of L2 Oral Proficiency using Speech Large Language Models](https://mlsanity.com/paper/2505.21148/) (2025)
- [Are Models Biased on Text without Gender-related Language?](https://mlsanity.com/paper/2405.00588/) (2024)
- [GSQA: An End-to-End Model for Generative Spoken Question Answering](https://mlsanity.com/paper/2312.09781/) (2023)
- [B-score: Detecting biases in large language models using response history](https://mlsanity.com/paper/2505.18545/) (2025)
- [Intelligibility of Text-to-Speech Systems for Mathematical Expressions](https://mlsanity.com/paper/2506.11086/) (2025)
- [ConversAR: Exploring Embodied LLM-Powered Group Conversations in Augmented Reality for Second Language Learners](https://mlsanity.com/paper/2505.24000/) (2025)

## Built on by (most cited first, 4 of 4)

- [Do LLM Decoders Listen Fairly? Benchmarking How Language Model Priors Shape Bias in Speech Recognition](https://mlsanity.com/paper/2604.21276/) (2026)
- [The Voice Behind the Words: Quantifying Intersectional Bias in SpeechLLMs](https://mlsanity.com/paper/2603.16941/) (2026)
- [A Cross Community Agenda for Speech AI](https://mlsanity.com/paper/2609.13168/) (2026)
- [Fairness Beyond a Single Run: Training-Seed Variability in Speech LLM Adaptation](https://mlsanity.com/paper/2609.38976/) (2026)

## Abstract (arXiv)

This work presents the first systematic investigation of speech bias in multilingual MLLMs. We construct and release the BiasInEar dataset, a speech-augmented benchmark based on Global MMLU Lite, spanning English, Chinese, and Korean, balanced by gender and accent, and totaling 70.8 hours (≈4,249 minutes) of speech with 11,200 questions. Using four complementary metrics (accuracy, entropy, APES, and Fleiss' kappa), we evaluate nine representative models under linguistic (language and accent), demographic (gender), and structural (option order) perturbations. Our findings reveal that MLLMs are relatively robust to demographic factors but highly sensitive to language and option order, suggesting that speech can amplify existing structural biases. Moreover, architectural design and reasoning strategy substantially affect robustness across languages. Overall, this study establishes a unified framework for assessing fairness and robustness in speech-integrated LLMs, bridging the gap between text- and speech-based evaluation. The resources can be found at https://github.com/ntunlplab/BiasInEar.