README.md

---
license: other
license_name: usc-research
license_link: LICENSE
language:
- en
pretty_name: VoxParadox
task_categories:
- audio-classification
- question-answering
tags:
- audio
- speech
- paralinguistic
- benchmark
- adversarial
- audio-llm
size_categories:
- 1K<n<10K
configs:
- config_name: default
  data_files:
  - split: test
    path: "audio/*.wav"
---

<div align="center">

<img src="assets/voxparadox_logo_full.png" alt="VoxParadox" width="360">

**Do Audio LLMs Listen or Read? Analyzing and Mitigating Paralinguistic Failures with VoxParadox**

Jiacheng Pang<sup>\*</sup>, Ashutosh Chaubey<sup>\*</sup>, Mohammad Soleymani \
University of Southern California \
<sub><sup>*</sup> Equal contribution</sub>

[![ICML 2026](https://img.shields.io/badge/ICML-2026-1d4ed8.svg)](https://icml.cc/Conferences/2026)
[![Paper](https://img.shields.io/badge/Paper-arXiv-AD1C18.svg)](https://arxiv.org/abs/2605.27772)
[![Project Page](https://img.shields.io/badge/Project-Page-0EA5E9.svg)](https://voxparadox.github.io/)
[![Code](https://img.shields.io/badge/GitHub-ihp--lab%2FVoxParadox-181717.svg?logo=github)](https://github.com/ihp-lab/VoxParadox)
[![Models](https://img.shields.io/badge/🤗%20Models-coming%20soon-FFD21E.svg)](#)
[![License](https://img.shields.io/badge/License-USC%20Research-228B22.svg)](LICENSE)

</div>

---

An adversarial speech QA benchmark for **paralinguistic understanding** in
Audio LLMs. Each example is built around a controlled linguistic–acoustic
contradiction: the transcript explicitly asserts an incorrect paralinguistic
attribute, while the audio reliably conveys the correct one. Models that
defer to transcript content are misled; models that listen are not.

**2,000 MCQs** across **10 paralinguistic tasks** (200 each).

## Quick start

```python
from datasets import load_dataset
ds = load_dataset("IHP-Lab/VoxParadox", split="test")
print(ds[0])           # includes `audio` (decoded), `question`, `choice_a..d`,
                       # `answer_gt`, `adversarial_labels`, `task_name`, `id`
```

## Tasks

`y_true` (audio) and `y_adv` (transcript) are disjoint by construction.

| Task (`task_name`) | Acoustic attribute the model must recover |
|---|---|
| `age_prediction` | Speaker's age group |
| `gender_prediction` | Speaker's gender |
| `emotion_recognition` | Emotion conveyed by delivery (high-contrast pairs) |
| `intonation_perception` | Rising vs. falling intonation |
| `speaker_identity_recognition` | Which segment shares a speaker with a queried segment |
| `total_speaker_counting` | Number of distinct speakers |
| `pitch_comparison` | Ordering of three segments by pitch |
| `volume_comparison` | Ordering of three segments by loudness |
| `speed_comparison` | Ordering of three segments by speaking rate |
| `vocal_range_comparison` | Ordering of three segments by pitch range |

## File layout

```
.
├── metadata.jsonl    # one record per example (loaded by `datasets`)
├── voxparadox.json   # same content as JSON array (for direct inspection)
├── audio/            # 2,000 wav files
├── eval.py           # evaluation script (GT accuracy + ALA)
├── assets/           # logo
└── LICENSE           # USC Research License
```

## Record schema

| Field | Type | Description |
|---|---|---|
| `id` | string | `{task_name}__{N}`, with `N` running 0–199 within each task. |
| `task_name` | string | One of the 10 tasks above. |
| `file_name` / `audio_path` | string | Path to the audio clip, relative to this directory. |
| `question` | string | The MCQ question prompt. |
| `choice_a` / `choice_b` / `choice_c` / `choice_d` | string | The four answer options. |
| `answer_gt` | string | The acoustic ground-truth `y_true` (one of the four choices). |
| `adversarial_labels` | list[string] | The transcript-implied label(s) `y_adv`. Single-element for most tasks; 2 elements for the four `*_comparison` tasks. |

## Evaluation

Two complementary metrics:

* **GT Accuracy** — fraction matching `answer_gt`. Higher is better; reflects use of acoustic evidence.
* **Adversarial-Label Agreement (ALA)** — fraction matching any string in `adversarial_labels`. Higher ALA means more transcript-following.

Run `eval.py` on a JSONL of model predictions (one record per line, fields `id` and `response`):

```bash
python eval.py --predictions preds.jsonl
```

Example prediction record:
```json
{"id": "age_prediction__0", "response": "(C) Elderly adult."}
```

The script parses A/B/C/D from the response (letter-first, then choice-text
fallback), prints per-task and overall GT/ALA, and optionally writes a JSON
report with `--report report.json`.

## License

Released under the **USC Research License** (research and non-profit use only;
commercial use requires a separate license). See [`LICENSE`](LICENSE) for the
full text.

Audio was synthesized via commercial TTS engines (ElevenLabs, GPT-4o,
Microsoft Azure); any commercial reuse of the audio is additionally subject
to those vendors' terms of service.

## Citation
```bibtex
@inproceedings{pang2026voxparadox,
  title     = {Do Audio LLMs Listen or Read? Analyzing and Mitigating Paralinguistic Failures with VoxParadox},
  author    = {Pang, Jiacheng and Chaubey, Ashutosh and Soleymani, Mohammad},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2026}
}
```