README.md

---
configs:
- config_name: zh_single
  data_files:
  - split: test
    path: data/zh_single/test.parquet
- config_name: zh_multi
  data_files:
  - split: test
    path: data/zh_multi/test.parquet
- config_name: en_single
  data_files:
  - split: test
    path: data/en_single/test.parquet
- config_name: en_multi
  data_files:
  - split: test
    path: data/en_multi/test.parquet
language:
- zh
- en
license: cc-by-nc-sa-4.0
task_categories:
- text-to-speech
tags:
- nonverbal-vocalization
- expressive-tts
- benchmark
- speech-synthesis
pretty_name: NV-Bench
size_categories:
- 1K<n<10K
---

# NV-Bench: Benchmarking Nonverbal Vocalization Synthesis in Expressive Text-to-Speech Models

[![Project Page](https://img.shields.io/badge/Project-Page-blue)](https://charlesnii.github.io/nvbench.github.io/)
[![GitHub](https://img.shields.io/badge/GitHub-Code-black)](https://github.com/AmphionTeam/NV-Bench)

## Abstract

While recent text-to-speech (TTS) systems increasingly integrate nonverbal vocalizations (NVVs), their evaluation lacks standardized metrics and reliable ground truth references. To bridge this gap, we propose **NV-Bench**, the first benchmark grounded in a functional taxonomy that treats NVVs as communicative acts rather than acoustic artifacts. NV-Bench comprises **1,651 multilingual, in-the-wild utterances** with paired human reference audio, balanced across **14 categories**. We introduce a dual-dimensional evaluation protocol:

1. **Instruction Alignment** — utilizes our proposed Paralinguistic Character Error Rate (PCER) to assess controllability.
2. **Acoustic Fidelity** — quantifies the distributional gap between synthesized and real speech.

Experimental results demonstrate a strong correlation between our objective metrics and human perception, establishing NV-Bench as a standardized evaluation framework.

## Dataset Overview

### Subsets

| Subset | Language | Description | Label Type |
|---|---|---|---|
| `zh_single` | Chinese | Single nonverbal event per utterance | Single-label |
| `zh_multi` | Chinese | Multiple nonverbal events per utterance | Multi-label |
| `en_single` | English | Single nonverbal event per utterance | Single-label |
| `en_multi` | English | Multiple nonverbal events per utterance | Multi-label |

- **Single-label Subset**: Strictly balanced — exactly one NVV event per utterance (50 samples per category). Isolates fundamental generation capabilities.
- **Multi-label Subset**: Challenging utterances with 2+ NVV events — tests robustness under dense paralinguistic conditions with relative balance.

### Data Fields

| Field | Type | Description |
|---|---|---|
| `text` | `string` | Target text with inline NVV tags (e.g. `[Laughter]`, `[Cough]`) |
| `prompt_text` | `string` | Prompt text for zero-shot speaker cloning |
| `category` | `string` | NVV category label (one of 14 categories) |
| `type` | `string` | Subset identifier (`zh_single`, `zh_multi`, `en_single`, `en_multi`) |
| `wav` | `Audio` | Ground-truth reference audio (MP3) |
| `prompt_wav` | `Audio` | Speaker prompt audio for zero-shot cloning (MP3) |

### Functional Taxonomy

NVVs are organized into **three functional levels** based on communicative intent:

| Name | Description | Categories |
|---|---|---|
| Vegetative Sounds | Biological reflexes grounding speech in physical realism | `[Cough]`, `[Sigh]`, `[Breathing]` |
| Affect Bursts | Valenced vocalizations conveying emotion or instant reactions | `[Laughter]`, `[Surprise-ah]`, `[Surprise-oh]`, `[Dissatisfaction-hnn]`|
| Conversational Grunts | Interaction-management cues — filled pauses and prosodic particles | `[Uhm]`, `[Confirmation-en]`, `[Question-ei]`, `[Question-ah]`, `[Question-en]`, `[Question-oh]`, `[Question-huh]` |

## Usage

```python
from datasets import load_dataset

# Load a specific subset
dataset = load_dataset("AnonyData/NV-Bench", "zh_single", split="test")

# Access a sample
sample = dataset[0]
print(sample["text"])          # Target text with NVV tags
print(sample["category"])      # NVV category
print(sample["wav"])           # Ground-truth audio
print(sample["prompt_wav"])    # Speaker prompt audio
print(sample["prompt_text"])   # Prompt text
```

## Evaluation Protocol

### Instruction Alignment
Measures whether the model can generate the specified NVV events at the correct positions.

| Metric | Description |
|---|---|
| **CER** | Character Error Rate |
| **PCER** | Paralinguistic Character Error Rate |
| **OCER** | Overall Character Error Rate |

### Acoustic Fidelity
Measures how realistic synthesized speech sounds compared to real recordings.

| Metric | Description |
|---|---|
| **FAD / FD / KL** | Distribution distance metrics |
| **SIM** | Speaker similarity |
| **DNSMOS** | Perceptual quality score |

## Pipeline

1. **Data Processing** — 565K clips (~1,560 hrs) filtered via Emilia-Pipeline & MiMo-Audio for single-speaker verification.
2. **Multi-lingual NVASR** — SenseVoice-Small fine-tuned on 6 datasets with unified label taxonomy.
3. **Human Verification** —  1,651 prompt-GT pairs (7.9 hrs).

## Citation

```bibtex
@article{ni2026nv,
  title={NV-Bench: Benchmark of Nonverbal Vocalization Synthesis for Expressive Text-to-Speech Generation},
  author={Ni, Qinke and Liao, Huan and Chen, Dekun and Wang, Yuxiang and Wu, Zhizheng},
  journal={arXiv preprint arXiv:2603.15352},
  year={2026}
}
```

## License

This dataset is released under the [CC BY NC SA 4.0](https://creativecommons.org/licenses/by-nc-sa/4.0/) license.