An International Phonetic Alphabet lexicon containing 6.9 million utterances across more than 350 languages. The speech files were converted to IPA using the neurlang/ipa-whisper-medium model and postprocessed for single-word headwords. The dataset was created by neurlang and last updated on May 4, 2026.
Use Cases
- Train phonetic transcription models based on the IPA-converted speech files.
- Analyze cross-language phonetic patterns based on the 350+ language coverage.
- Build multilingual pronunciation dictionaries based on the single-word IPA records.
- Benchmark speech-to-text systems using the large-scale IPA lexicon.
- Study language phonology using the structured IPA representations.
Strengths
- Contains 6.9 million utterances, providing a large-scale phonetic resource.
- Covers more than 350 languages, offering broad multilingual coverage.
- Utilizes a specific model (ipa-whisper-medium) for consistent IPA conversion.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Description metadata is limited; actual data quality requires manual inspection after download.
- Row count and file formats are unknown, which may limit suitability assessment.
Provenance
- Source
- neurlang
- Collection Method
- Speech files converted to IPA using the neurlang/ipa-whisper-medium model.
- Freshness
- Last updated 2026-05-04 18:56:33; freshness should be verified.