Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html>
<h"... is not valid JSON
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
IPA Phonetic Lexicon (6.9M words)
- This is an International Phonetic Language lexicon containing 6.9 million utterances across 350+ languages.
- All speech files were converted into IPA using our neurlang/ipa-whisper-medium model.
- Postprocessing to join multiword IPA into single-word IPA record for single-word headwords was performed for appropriate languages.
Global Map
Stats
- Total words: 6908075
- Countries: 211
- Languages: 356
- Unique Speakers: 81756
- Gender Ratio M/F: 4802600/2083464
Data Format
JSONL/ZST
{
"language": "en",
"headword": "café",
"query_word": "caf%C3%A9",
"votes": 6,
"origin": ["user", "male", "united kingdom"],
"download_url": "...",
"is_ogg": false,
"id": 55272,
"ipa": "kafˈe"
}
Quality scoring (votes)
| words | votes |
|---|---|
| 6470891 | 0 |
| 324214 | 1 |
| 44413 | 2 |
| 14218 | 3 |
| 6491 | 4 |
| 3552 | 5 |
| 2143 | 6 |
| 1392 | 7 |
| 956 | 8 |
| 635 | 9 |
| 1762 | 10+ |
- Votes represent community confidence in pronunciation accuracy
- Recommended filter: votes ≥ 1 for high-quality subset
- Negative votes are also present (not shown above)
- In the future other quality metrics (CER w.r.t. grammar) will be added
⚠️ Limitations
- Speaker distribution is highly uneven across countries
- Some languages are overrepresented (e.g. English, Spanish)
- Country labels/languages may be inacurrate for some users (e.g. Antartica)
- Audio quality varies by contributor
- IPA generation is model-derived (not manually verified)
- Regional dialect distinctions may be blurred
- Downloads last month
- 398