Every word deserves an audible place
Audio is part of the word—not a link maze.
Every safe exact match now plays beside the first visible headword. Counts below come from the complete current encyclopedia denominator and an audited, revision-pinned Hugging Face inventory.
768,721exact public headwords
282,942headwords with safe public audio
36.81%exact-headword coverage
485,779exact headwords still missing
197,336clips held from public wiring
394,068audio files in the live public HF repo
What “safe public audio” means
Each visible clip has an exact headword join, exists at an audited immutable HF revision, passed existing technical and same-asset ASR gates, and is not a known cross-headword waveform conflict or non-atomic label.
Synthetic CAB audio. Technical and same-asset ASR validation is not fluent-speaker pronunciation certification; human review remains open.
- Public OPUS files on HF
- 231,374
- Public WAV files on HF
- 162,694
- Eligible exact clips
- 282,939
- Model / checkpoint license
- facebook/mms-tts-cab · CC-BY-NC-4.0