Mozilla makes people read sentences aloud to train their open voice recognition system: https://voice.mozilla.org/

Would be nice to understand if they can use text for which we'd need a recording, such as Wiktionary entries/citations.

Nemo