Audio-Language Model Evaluation

Overview

The field measures how well models use audio to answer questions, follow spoken instructions, describe sounds, and reason about events over time. Researchers build benchmarks for speech, music, environmental sound, long recordings, multilingual settings, and safety, using multiple-choice and open-ended tasks alongside human ratings. Controlled tests—such as removing or altering audio, changing answer order, or making speech conflict with text—help reveal shortcuts, bias, and whether a model’s answer is actually grounded in sound; uncertainty estimates and model-based judges can make evaluation more scalable. A central challenge is that strong benchmark scores may coexist with weak acoustic grounding, temporal reasoning, and robustness to changes in wording, speakers, or recording conditions.

Latest papers

225 papers23 papers in the last four weeks, down 38% on the four weeks before. 0.2% of all new papers.

Feb 19, 2026

The Cascade Equivalence Hypothesis: When Do Speech LLMs Behave Like ASR→\rightarrowLLM Pipelines?

Speech LLMs are widely understood to be better than ASR→\rightarrowLLM cascades since they have access to the audio directly, and not just the transcript. In this paper, we present an evaluation methodology and a mechanistic interpretation of the observed behavior of speech LLMs. First, we introduce matched-backbone testing which separates out the behavior of the speech LLM from the reasoning capabilities of the underlying LLM. Second, we provide a mechanistic analysis of speech LLMs using logit lens and LEACE and show the literal transcript emerging from the LLM's hidden states and that text representations are causally necessary. We also show that in most deployed use cases, current speech LLMs are expensive cascades, and under noise, they are worse ones, with clean-condition advantages reversing by up to 7.6% at 0dB.
Feb 17, 2026

MAEB: Massive Audio Embedding Benchmark

We introduce the Massive Audio Embedding Benchmark (MAEB), a large-scale benchmark covering 30 tasks across speech, music, environmental sounds, and cross-modal audio-text reasoning in 100+ languages. We evaluate 50+ models and find that no single model dominates across all tasks: contrastive audio-text models excel at environmental sound classification (e.g., ESC50) but score near random on multilingual speech tasks (e.g., SIB-FLEURS), while speech-pretrained models show the opposite pattern. Clustering remains challenging for all models, with even the best-performing model achieving only modest results. We observe that models excelling on acoustic understanding often perform poorly on linguistic tasks, and vice versa. We also show that the performance of audio encoders on MAEB correlates highly with their performance when used in audio large language models. MAEB is derived from MAEB+, a collection of 98 tasks. MAEB is designed to maintain task diversity while reducing evaluation cost, and it integrates into the MTEB ecosystem for unified evaluation across text, image, and audio modalities. We release MAEB and all 98 tasks along with code and a leaderboard at https://github.com/embeddings-benchmark/mteb.
Feb 4, 2026

RA-QA: A Benchmarking System for Respiratory Audio Question Answering Under Real-World Heterogeneity

As conversational multimodal AI tools are increasingly adopted to process patient data for health assessment, robust benchmarks are needed to measure progress and expose failure modes under realistic conditions. Despite the importance of respiratory audio for mobile health screening, respiratory audio question answering remains underexplored, with existing studies evaluated narrowly and lacking real-world heterogeneity across modalities, devices, and question types. We hence introduce the Respiratory-Audio Question-Answering (RA-QA) benchmark, including a standardized data generation pipeline, a comprehensive multimodal QA collection, and a unified evaluation protocol. RA-QA harmonizes public RA datasets into a collection of 9 million format-diverse QA pairs covering diagnostic and contextual attributes. We benchmark general audio-language models as well as domain-specific architectures, establishing reproducible reference points and showing how current approaches fail under heterogeneity.
Jan 30, 2026

The Alignment Curse: Modality Alignment Supercharges Audio Attacks via Text Transfer

Recent advances in end-to-end trained omni-models have substantially improved audio capabilities by strengthening text-audio modality alignment. However, whether such alignment inadvertently facilitates the transfer of safety vulnerabilities across modalities remains underexplored. This question is critical as text-based jailbreak attacks are considerably more mature than audio-based ones; if they transfer systematically, current audio safety evaluations may underestimate risks originating from the text modality. In this paper, we introduce the Alignment Curse, a formally characterized and empirically validated principle showing that stronger modality alignment enables more effective transfer of attacks from text to audio, revealing a fundamental tension between capability and safety. Motivated by this principle, we conduct a comprehensive black-box evaluation of three attack categories on recent omni-models (e.g., Qwen2.5-Omni, Qwen3-Omni): text attacks, text-transferred audio attacks, and audio attacks. We find that text-transferred audio attacks perform comparably to, and often better than, audio-based attacks, exhibiting a clear advantage under audio-only access. This suggests that text-based vulnerabilities play a pivotal role in shaping audio safety risks. Finally, we empirically analyze the relationship between modality alignment and transfer effectiveness across attack methods and models, observing consistent support for the Alignment Curse: tighter modality alignment leads to more effective cross-modality attack transfer.
Jan 29, 2026

SONIC-O1: A Real-World Benchmark for Evaluating Multimodal Large Language Models on Audio-Video Understanding

Multimodal Large Language Models (MLLMs) are a major focus of recent AI research. However, most prior work focuses on static image understanding, while their ability to process sequential audio-video data remains underexplored. This gap highlights the need for a high-quality benchmark to systematically evaluate MLLM performance in a real-world setting. We introduce SONIC-O1, a comprehensive, fully human-verified benchmark of 60 hours (231 clips) spanning 13 real-world conversational domains with 4,958 annotations and demographic metadata. SONIC-O1 evaluates three capabilities: open-ended summarization, multiple-choice question (MCQ) answering, and temporal localization with supporting rationales (reasoning). Across closed- and open-source models, we find that the MCQ accuracy shows the smallest gap between model families, but the best closed-source model outperforms the best open-source model by 22.6% on temporal localization. We further observe accuracy gaps of up to 21.4% on temporal localization across demographic groups, indicating persistent disparities in model behaviour. SONIC-O1 provides an open evaluation suite for temporally grounded and demographically robust multimodal understanding. SONIC-O1 is publicly available for research: Project page (https://vectorinstitute.github.io/sonic-o1/), Dataset (https://huggingface.co/datasets/vector-institute/sonic-o1), GitHub (https://github.com/vectorinstitute/sonic-o1), Leaderboard (https://huggingface.co/spaces/vector-institute/sonic-o1-leaderboard).
Jan 6, 2026

The World is Not Mono: Enabling Spatial Understanding in Large Audio-Language Models

Large audio-language models have made rapid progress in recognizing what is present in an audio clip, but spatial audio-language understanding still lacks a clear task interface. A model must also decide where sound events occur, which semantic and spatial attributes belong to the same auditory object, how multiple objects are arranged, and whether a scene-level answer is physically plausible. We formalize this capability as audio scene analysis (ASA), a three-level problem spanning atomic perception, relational integration, and cognitive reasoning. We propose The World is Not Mono (TWNM), a framework that equips audio-language models with explicit spatial evidence. TWNM uses physically grounded First-Order Ambisonics (FOA) simulation for controllable supervision, learns slot-regularized spatial representations from multichannel audio, fuses them with semantic audio features, and trains with a progressive curriculum ending in preference optimization over metadata-derived correct answers and auxiliary format/evidence rewards. To operationalize ASA, we build a controlled benchmark from scene metadata, covering localization, attribute binding, spatial comparison, scene abduction, and counterfactual reasoning. On this benchmark, TWNM achieves 70.8% overall accuracy, 66.4% on spatial-family tasks, and 79.76% on mixed L3\mathcal{L}_3 scene-level QA under exact multiple-choice question answering (MCQA) scoring. We also audit monaural and binaural reference systems as diagnostic references with explicit audit labels, since they differ in spatial input, training interface, and output format. The supported claim is that a clearly defined ASA hierarchy, FOA-conditioned spatial representations, and metadata-grounded training enable controlled, auditable spatial audio-language reasoning, with STARSS23 providing a limited real-recording diagnostic.
Nov 28, 2025

ORCA: Open-ended Response Correctness Assessment for Audio Question Answering

Reliable assessment of the abilities of large audio language models (LALMs) is essential to advancing the state of the art. As benchmarks rapidly evolve to incorporate complex reasoning and subjective tasks, they increasingly necessitate open-ended responses from LALMs. We present Open-ended Response Correctness Assessment (ORCA) -- a reliable and lightweight model-based approach for answer correctness and disagreement modeling. We employ a three-stage annotation pipeline combining human judgment, structured feedback, and human-AI correction, yielding 9,663 annotations across 3,699 question-answer pairs from 15 LALMs on three audio understanding and reasoning benchmarks (achieving a Krippendorff's alpha of 0.82). Our experiments employing curriculum learning show that ORCA models achieve a Spearman correlation of 0.91 with average human correctness ratings on seen benchmarks and generalize to unseen benchmarks with a score of 0.85, outperforming several LLM judge baselines including Gemini 2.5 Flash. Furthermore, we demonstrate that ORCA's predicted variance correlates strongly with human disagreement, allowing it to effectively identify problematic benchmark items.
Nov 20, 2025

Revisiting Audio-language Pretraining for Learning General-purpose Audio Representation

Audio-language pretraining (ALP) holds promise for learning general-purpose audio representation, yet remains underexplored. Crucially, there is no consensus on whether audio-language models can build effective general-purpose audio encoders, nor a systematic understanding of how pretraining objectives behave across diverse tasks and scales. We identify three key barriers: limited scale of audio-text corpora, limited coverage of audio attributes in existing caption corpora, and lack of systematic exploration and evaluation. To fill this gap, we present the first principled empirical study of ALP. We first introduce CaptionStew, a 10.7M caption dataset aggregating open-source audio-text corpora across multiple domains and captioning focuses. We then conduct the first comprehensive evaluation comparing contrastive and captioning objectives for learning audio representation across speech, music, and environmental sound tasks. Our results not only demonstrate that ALP yields competitive, transferable representations, but reveal critical trade-offs: contrastive learning offers superior data efficiency, while captioning exhibits better scalability. Furthermore, we find that the benefits of supervised initialization often diminish at larger scales, challenging common practices. By grounding these claims in empirical evidence, we establish a viable pathway toward general-purpose audio representation learning, guiding future research.
Nov 3, 2025

Speech-DRAME: A Framework for Human-Aligned Benchmarks in Speech Role-Play

Evaluating spoken role-play requires judging not only linguistic content but also prosody, emotional expression, character embodiment, and contextual fit. Existing automatic evaluation methods often collapse these aspects into coarse scores and are evaluated primarily on synthesized speech. We introduce Speech-DRAME, a framework comprising Speech-DRAME-EvalBench, a bilingual, human-annotated benchmark with contrastive role-, context-, and delivery-mismatch conditions, and DRAME-Eval, a fine-tuned audio-language evaluator. The benchmark supports two complementary protocols: top-down archetype evaluation and bottom-up realism evaluation, with predominantly synthesized benchmark-distribution data and a held-out stress test consisting entirely of authentic human acting. On the benchmark-distribution sets, DRAME-Eval achieves pooled Pearson correlations of 0.629 for archetype evaluation and 0.602 for realism evaluation, significantly outperforming the evaluated zero- and few-shot audio-language judges; modality controls show that its judgments depend on both speech and textual context. This alignment does not transfer to the authentic-speech test set, where DRAME-Eval obtains an average correlation of -0.142 and a within-speaker correlation of +0.029, compared with matched human single-rater agreement of +0.34. These results identify a substantial synthetic-to-authentic transfer gap and establish Speech-DRAME as a resource for studying and improving evaluation of acting differences among real human performances.
Oct 13, 2025

VCB Bench: An Evaluation Benchmark for Audio-Grounded Large Language Model Conversational Agents

Recent advances in large audio language models (LALMs) have greatly enhanced multimodal conversational systems. However, existing benchmarks remain limited -- they are mainly English-centric, rely on synthetic speech, and lack comprehensive, discriminative evaluation across multiple dimensions. To address these gaps, we present Voice Chat Bot Bench (VCB Bench) -- a high-quality Chinese benchmark built entirely on real human speech. VCB Bench evaluates LALMs from three complementary perspectives: instruction following (including speech-level control beyond text commands), knowledge understanding (general knowledge, reasoning, and daily dialogue), and robustness (stability under perturbations in content, environment, and speaker traits). Experiments on representative LALMs reveal notable performance gaps and highlight future directions for improvement. VCB Bench provides a reproducible and fine-grained evaluation framework, offering standardized methodology and practical insights for advancing Chinese voice conversational models.
Oct 6, 2025

Robustness assessment of large audio language models in multiple-choice evaluation

Recent advances in large audio language models (LALMs) have primarily been assessed using a multiple-choice question answering (MCQA) framework. However, subtle changes, such as shifting the order of choices, result in substantially different results. Existing MCQA frameworks do not account for this variability and report a single accuracy number per benchmark or category. We dive into the MCQA evaluation framework and conduct a systematic study spanning three benchmarks (MMAU, MMAR and MMSU) and four models: Audio Flamingo 2, Audio Flamingo 3, Qwen2.5-Omni-7B-Instruct, and Kimi-Audio-7B-Instruct. Our findings indicate that models are sensitive not only to the ordering of choices, but also to the paraphrasing of the question and the choices. Finally, we propose a simpler evaluation protocol and metric that account for subtle variations and provide a more detailed evaluation report of LALMs within the MCQA framework.
Oct 1, 2025

Hearing the Order: Investigating Position Bias in Large Audio-Language Models

Large audio-language models (LALMs) are often used in tasks that involve reasoning over ordered options. An open question is whether their predictions are influenced by the order of answer choices, which would indicate a form of position bias and undermine their reliability. In this paper, we identify and analyze this problem in LALMs. We demonstrate that no model is immune to this bias through extensive experiments on six LALMs across three widely used benchmarks and their spoken counterparts. Shuffling the order of answer options can cause performance fluctuations of up to 24% and even change model rankings, raising concerns about the reliability of current evaluation practices. We also study permutation-based strategies and show that they can mitigate bias in most cases. Our work represents the first systematic investigation of this issue in LALMs, and we hope it raises awareness and motivates further research in this direction.
Sep 18, 2025

Evaluating Multimodal Large Language Models on Spoken Sarcasm Detection

Sarcasm detection remains challenging because sarcastic intent often depends on information across text, speech, and vision. We systematically evaluate large language models (LLMs) and multimodal LLMs (MLLMs) for sarcasm detection on English (MUStARD++) and Chinese (MCSD), considering both feature-extraction settings and direct classification. In the feature-extraction settings, audio provides the strongest unimodal performance on both datasets, while multimodal fusion shows dataset-dependent benefits. In direct classification, Qwen2.5-Omni achieves the strongest zero-shot performance among the evaluated models, with further substantial gains after fine-tuning. Error analysis further shows that additional modalities can both correct and introduce errors, highlighting the need for selective multimodal integration.
Sep 9, 2025

AU-Harness: An Open-Source Toolkit for Efficient and Unified Evaluation of AUdio-LLMs

Large Audio Language Models (LALMs) are rapidly advancing, but evaluating them remains challenging due to inefficient and nonstandardized toolkits that limit fair comparison and systematic assessment. Existing evaluation frameworks exhibit three critical limitations: (1) slow and inefficient processing pipeline that bottlenecks large-scale studies, (2) inadequate multi-turn dialogue support, leaving fundamental questions about cross-turn context integration and performance dynamics over extended conversations in LALMs unanswered; and (3) the absence of unified and scalable evaluation framework capable of keeping pace with the rapid growth of both LALMs and audio benchmarks. To address these issues, we introduce AU-Harness, an efficient and comprehensive evaluation framework for LALMs. Our empirical studies demonstrate that AU-Harness achieves 77% reduction in evaluation processing time over existing evaluation toolkits, enabling large-scale LALM evaluations previously considered impractical. We provide standardized evaluation protocols and flexible configurations for fair model comparison across diverse scenarios. AU-Harness unlocks a range of in-depth analyses difficult to conduct without a unified foundation, including multi-turn dialogue dynamics, enabling the study of true audio reasoning capabilities in existing LALMs. AU-Harness provides both practical evaluation tools and insights into model limitations, advancing systematic LALM development.
Aug 28, 2025

WoW-Bench: Evaluating Fine-Grained Acoustic Perception in Audio-Language Models via Marine Mammal Vocalizations

Large audio-language models (LALMs) extend language understanding into the auditory domain, yet their ability to perform low-level listening, such as pitch and duration detection, remains underexplored. However, low-level listening is critical for real-world, out-of-distribution tasks where models must reason about unfamiliar sounds based on fine-grained acoustic cues. To address this gap, we introduce the World-of-Whale benchmark (WoW-Bench) to evaluate low-level auditory perception and cognition using marine mammal vocalizations. We use marine mammal vocalizations as out-of-distribution sound events to better assess models' low-level listening and so that the models do not rely on prior knowledge of the sound events. WoW-bench is composed of a Perception benchmark for categorizing novel sounds and a Cognition benchmark, inspired by Bloom's taxonomy, to assess the abilities to remember, understand, apply, and analyze sound events. For the Cognition benchmark, we additionally introduce distractor questions to evaluate whether models are truly solving problems through listening rather than relying on other heuristics. Experiments with state-of-the-art LALMs show performance far below human levels, indicating a need for stronger auditory grounding in LALMs.