Meorav: speech synthesis for English-Hebrew code-switching

Hebrish, my English–Hebrew speech project, now has a voice. Meorav is a text-to-speech model built for code-switching: give it a sentence of English with Hebrew words in it, written in Latin letters, and it reads the whole thing aloud, Hebrew words included, in one voice. Version 2.2, released on 9 November 2026, is the first one I'm sharing as something to use rather than a proof of concept. (The project's own launch post is on hebrish.app; this is my version of it.)

And yes, it has an Irish accent.

About the accent

Meorav is trained on one voice: mine. I'm Irish and I live in Jerusalem, so the model reads English like an Irishman and Hebrew like someone who has spent a lot of time in a lot of offices waiting for a lot of forms. That's what it learned, and that's what it gives back.

I didn't try to remove the accent, and this release makes no claim about it. Nor is it a claim of native Hebrew: I'm not a native speaker, and Meorav doesn't speak Hebrew like one either (a later version may also try for a more neutral English). What it does demonstrate is the code-switching case: English with Hebrew words dropped in, the way English speakers living in Israel actually talk. Most English text-to-speech voices turn Misrad Hapnim, teudat zehut or acharei hachagim into something a tourist would say from a phrasebook: wrong stress, a ch that comes out as a k, Hebrew read as if it were English spelling. Meorav says them the way an English speaker who lives here says them in the middle of a sentence: as Hebrew words, not as English spelling, without a pause or a change of voice. It doesn't always get them right; listen and judge for yourself.

Listen

The same sentence twice. Before is the base model, untuned, copying my voice from a single recording. After is Meorav. Listen to the Hebrew words.

1. I spent the whole morning at Misrad Hapnim, and I still don't have the teudat zehut.

2. Everyone here says everything will happen acharei hachagim. I used to find it frustrating, but honestly, after a few years you start saying it yourself.

3. The arnona bill came in twice this month, so I rang the iriya. After forty minutes on hold they told me, beseder, it's a mistake, just ignore the second one.

There are ten sentences in all, from a single line up to a 35-second passage, on Meorav's page on my AI site, on hebrish.app and on Hugging Face. None was picked or edited.

A little flat

Listen to a few and you'll hear that Meorav reads rather evenly, like someone reading a sentence off a card. That's more or less what it was trained on: almost every take is a read prompt, and every take goes through the same compression and loudness levelling. The cheapest thing to try is sampling: the model samples its speech at a "temperature" of 0.9 by default, and a higher one lets it pick less likely, livelier phrasing. Here's the second sentence again at 1.1 and 1.3:

Every one of the ten samples is on the model page at both temperatures, so you can hear the difference for yourself. Longer-term, the fix is more expressive training audio, which I'm recording now.

What it was trained on

• 3,995 takes, about 5.5 hours, all me. 1,103 were recorded specifically for the voice model; the other 2,924 come from the recordings I made for the Hebrish speech recognition models, screened down to everyday code-switching sentences.

• Every take was cleaned with NVIDIA NemotronLabs RE-USE, then a podcast-style voice chain.

• The base is Qwen3-TTS-12Hz-1.7B-Base, fully fine-tuned (every weight updated) for 3 epochs.

Two earlier attempts at my voice, an Orpheus fine-tune and a zero-shot clone, both came out sounding American. (I put an earlier Meorav up against myself in a ten-minute conversation, if you want to hear how that went.)

Getting it

The weights are on Hugging Face at Hebrish/Meorav, by request, under CC BY-NC 4.0 (free for non-commercial use with attribution), with GGUF builds for running it on a device. There's also a beta Android app that runs it on the phone. The model page on hebrish.app has example code and hardware notes.

There's no benchmark for Meorav yet; I released it after listening to it. The samples are there so you can judge it for yourself, accent and all.