Large downloadThe model is about 660 MB. Your browser downloads it once, then keeps it. Best on a desktop or laptop in Chrome, Edge or Firefox; once loaded it uses about 2.5 GB of memory.
On a phone, the Hebrish app is the better way to try it: the same model, offline, as a keyboard.
Open the full page to try it
Here, inside Hugging Face's frame, your browser won't let the model use more than one processor core, and it would freeze while it starts. The full page runs it properly.
1. Load the model
Checking whether the model is already saved in this browser…
2. Say something
Press Record, read a sentence below (or your own), then press Stop. Up to 30 seconds.
What Jachnun heard
Sentences to read aloud
The Hebrew word should come out in Latin letters, spelt as shown. The Hebrew script is only a reference; the model never writes it.
- I need to go to the iriyaעירייה to pay my arnonaארנונה bill.
- Did you see where I put my teudat zehutתעודת זהות?
- Do you want ambaעמבה on your sabichסביח or just tahini?
- The office was a complete balaganבלגן after the move.
Every word it was taught: the lexicon.
How it compares
Public benchmark, clean subset: 84 clips, 172 Hebrew term occurrences, 139 distinct terms. Full results.
| Stock Parakeet v2 | Jachnun v1.0 | |
|---|---|---|
| Hebrew terms transcribed correctly | 32.6% | 73.3% |
| Terms never seen in training | 32.5% | 70.1% |
| Word error rate | 9.5% | 3.3% |
| Plain English word error rate | 2.7% | 2.2% |
Jachnun was trained on one speaker's voice. If you speak English with Hebrew in it, send us a few recordings and the next version learns yours too.
How this page works
It runs the int8 ONNX build of Jachnun v1.0, downloaded from Hebrish's own CDN, with parakeet.js and ONNX Runtime Web on your processor. The files are kept in your browser's storage; Remove the saved model deletes them.