belztjti/aaaa111223

VoxAging

VoxAging: Continuously Tracking Speaker Aging with a Large-Scale Longitudinal Dataset in English and Mandarin

📄 Paper: https://arxiv.org/pdf/2505.21445

---

Description

The VoxAging dataset is a large-scale longitudinal audio-visual corpus designed for studying long-term speaker aging and temporal variations in speech.
It consists of recordings from 293 speakers, including 226 English speakers (112 female, 114 male) and 67 Mandarin speakers (23 female, 44 male).

All data were sourced from YouTube and Bilibili. Speaker channels were manually filtered to ensure:

• consistent speaker identity,
• sufficient temporal coverage,
• high audio-visual quality.

---

Access

This dataset is gated. Access to this dataset is gated. By requesting access, users confirm that they have read and agree to the VoxAging Dataset Usage Agreement.

---

Intended Use

This dataset is intended for research purposes only, including but not limited to:

• speaker aging analysis,
• speaker recognition and verification,
• longitudinal speech modeling,
• cross-lingual speaker analysis.

Commercial use is not permitted.

---

Ethical Considerations

• All data were collected from publicly available online sources.
• The dataset is intended strictly for academic research.
• No personal identifying information beyond publicly available speaker content is provided.
• Users are expected to comply with relevant data protection and ethical research guidelines.
• Redistribution of the dataset or any derived content is prohibited.

---

Citation

If you use this dataset in your research, please cite:

@inproceedings{ai25_interspeech,
  title     = {VoxAging: Continuously Tracking Speaker Aging with a Large-Scale Longitudinal Dataset in English and Mandarin},
  author    = {Zhiqi Ai and Meixuan Bao and Zhiyong Chen and Zhi Yang and Xinnuo Li and Xinnuo Li and Shugong Xu},
  booktitle = {Interspeech 2025},
  year      = {2025},
  pages     = {3648--3652},
  doi       = {10.21437/Interspeech.2025-57},
}