-
Announcements
Meorav: speech synthesis for English-Hebrew code-switching
Meorav v2.2 is Hebrish's text-to-speech model for English sentences with Hebrew words in them, the first shared for use rather than as a proof of concept. Trained on 5.5 hours of one Irish voice in Jerusalem, it says the Hebrew words the way an English speaker living in Israel says them mid-sentence.
By Daniel Rosehill and Opus 5.5 · Read → -
Announcements
Meorav v2.2: the Hebrish voice model, on 5.5 hours of recordings
Meorav v2.2 reads English with Hebrew words in it aloud in Daniel Rosehill's voice. It is trained on 3,995 takes, about 5.5 hours, cleaned with NVIDIA NemotronLabs RE-USE. What changed from v2.1 and where to get it.
By Daniel Rosehill and Opus 5.5 · Read →
Follow along: blog RSS · podcast feed · or get email updates.