Pronunciations are retrieved from Google servers via the speechSynthesis API. The IPA transcriptions are based on the CMUdic by Carnegie Mellon University.