Loading datasets...
License:
CC0-1.0
Steward:
TidyVoice2026 ChallengeDataset ID:
cmu6xsi8201eenq07hnip96ct
Release Date: 9/18/2026
Format: WAV
Size: 641.76 GB
Tidy-M is the monolingual partition of TidyVoice, a speaker-verification corpus derived from Mozilla Common Voice and cleaned so that each identity really is one person. It holds 212,617 speakers who contributed in a single language across 81 languages, 5.6M read-speech utterances, 8,138 hours, WAV at 16 kHz. The split is speaker-disjoint: 141,623 speakers for training (speakers with more than four utterances) and 70,994 for testing, with at least five test speakers per language. The evaluation condition is intra-lingual, ~2.8M trials where both sides share the same language, released per language as well as in aggregate.
According to Mozilla’s usage rules, it is forbidden to use this dataset for speaker identification or for recovering a speaker’s identity.
According to Mozilla’s usage rules, it is forbidden to use this dataset for speaker identification or for recovering a speaker’s identity. The data MUST only be used for speaker verification tasks.
Intended Use
All rules and restrictions are the same as those of the original Mozilla Common Voice datasets.
Tidy-M is a large-scale, massively multilingual speech corpus curated for intra-lingual (same-language) speaker verification research. It is the monolingual partition of the TidyVoice dataset, derived from Mozilla Common Voice (MCV) after mitigating the speaker heterogeneity that affects MCV client IDs.
🌐 Project / models: https://github.com/areffarhadi/wespeaker/tree/master/examples/tidyvocie
| Metric | Training Set | Test Set | Total |
|---|---|---|---|
| Speakers | 141,623 | 70,994 | 212,617 |
| Languages | 81 | 81 | 81 |
| Utterances | 5.4M | 218K | 5,616,611 |
| Duration (hours) | 7,800 | 350 | 8,138 |
| Trial pairs | — | ~2.8M | ~2.8M |
| Domain | Read Speech | Read Speech | Read Speech |
Splits prioritize a large training set:
The Tidy-M evaluation condition is intra-lingual: both sides of every trial are in the same language.
| Trial type | Speaker | Language |
|---|---|---|
| Target | Same | Same |
| Non-target | Different | Same |
The full test trial list contains approximately 2.8 million pairs. Trials are also provided per language for all 81 languages, enabling fine-grained analysis of model performance across diverse linguistic contexts.
📥 Download: Tidy-M test trial pairs
Trial file format (WeSpeaker/VoxCeleb style):
1 id010001/en/utt_0001.wav id010001/en/utt_0007.wav
0 id010001/en/utt_0001.wav id010042/en/utt_0003.wav
If you use the Tidy-M dataset in your research, please cite:
@inproceedings{farhadipour2026tidyvoice,
title={TidyVoice: A Curated Multilingual Dataset for Speaker Verification Derived from Common Voice},
author={Farhadipour, Aref and Marquenie, Jan and Madikeri, Srikanth and Chodroff, Eleanor},
booktitle={ICASSP},
year={2026}
}
@article{zhang2025quantifying,
title={Quantifying and reducing speaker heterogeneity within the Common Voice corpus for phonetic analysis},
author={Zhang, Miao and Farhadipour, Aref and Baker, Annie and Ma, Jiachen and Pricop, Bogdan and Chodroff, Eleanor},
journal={arXiv preprint arXiv:2506.00733},
year={2025}
}
This dataset is derived from Mozilla Common Voice and is distributed under the same terms (CC-0). Please refer to the Mozilla Common Voice license for usage terms and conditions.