Automatic Speech Recognition
K2
ONNX
Hebrew
speech
asr
hebrew
streaming
zipformer
transducer
icefall
sherpa-onnx
proof-of-concept
Instructions to use Hebrish/Gefilte with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- K2
How to use Hebrish/Gefilte with K2:
# No code snippets available yet for this library. # To use this model, check the repository files and the library's documentation. # Want to help? PRs adding snippets are welcome at: # https://github.com/huggingface/huggingface.js
- Notebooks
- Google Colab
- Kaggle
|
Download README.md from Hebrish/Gefilte: direct link, hf CLI and curl.
- Browser
- Download file 5.28 kB
-
https://huggingface.co/Hebrish/Gefilte/resolve/main/README.md
- Command line
-
hf download hf://Hebrish/Gefilte/README.md
-
curl -L -o README.md https://huggingface.co/Hebrish/Gefilte/resolve/main/README.md
5.28 kB
| license: cc-by-4.0 | |
| language: | |
| - he | |
| pipeline_tag: automatic-speech-recognition | |
| base_model: Hebrish/Zipformer2-66M-Streaming-He | |
| tags: | |
| - speech | |
| - asr | |
| - hebrew | |
| - streaming | |
| - zipformer | |
| - transducer | |
| - icefall | |
| - k2 | |
| - sherpa-onnx | |
| - proof-of-concept | |
| # Gefilte (PoC-1) | |
| **Gefilte** is Daniel Rosehill's Hebrew speech-recognition fine-tune: a Hebrew model adapted to his own non-native (Anglo | |
| oleh) accent and the vocabulary he actually needs. The name is a joke: Daniel's other models are named after food, and | |
| an Ashkenazi immigrant's Hebrew deserved gefilte fish. | |
| > **Proof of concept.** PoC-1 was trained on **4.9 minutes** of one speaker reading Hebrew from nikud. Its only | |
| > purpose is to show that a few minutes of recordings can teach the base a word it has never heard. The test word is | |
| > Daniel's surname as he writes it in Hebrew, **专讜讝讛讬诇**. It is not a general improvement on the base, and it is not | |
| > recommended for anyone else's voice. It is not part of the Hebrish code-switching models. | |
| ## Base | |
| [`Hebrish/Zipformer2-66M-Streaming-He`](https://huggingface.co/Hebrish/Zipformer2-66M-Streaming-He): the Hebrish | |
| project's unchanged fork of [`ivrit-ai/zipformer2-66m-streaming-he`](https://huggingface.co/ivrit-ai/zipformer2-66m-streaming-he), | |
| **developed by Yakov Kolani** and trained on [ivrit.ai](https://www.ivrit.ai) Hebrew audio (CC BY 4.0). It is a 66M | |
| Zipformer2 streaming transducer (icefall / k2). Gefilte keeps its architecture, tokenizer (`bpe.model`, 1,493 tokens) and | |
| streaming configuration (chunk 64 frames = 1.28 s, left context 256 frames), and changes only the weights. | |
| ## Data | |
| Takes from the `takes` tab of [`danielrosehill/Gefilte-Hebrew-Sentences`](https://huggingface.co/datasets/danielrosehill/Gefilte-Hebrew-Sentences), | |
| recorded in Daniel's capture app. He reads each Hebrew sentence from a pointed (nikud) version, with | |
| [DictaBERT menaked](https://huggingface.co/dicta-il/dictabert-large-char-menaked) supplying the nikud. The prompts are | |
| graded Tatoeba sentences plus a priority set: his surname and spoken dev and AI Hebrew. Targets are the unpointed Hebrew | |
| `text`, with the punctuation the base vocabulary can write (`, . ? ! : ; " ( ) -`). | |
| | | Takes | Minutes | | |
| |---|---|---| | |
| | Trained on | 84 | 4.9 | | |
| | Held out (never trained on) | 8 | | | |
| The held-out takes are the dataset's test rows, the two newest takes containing 专讜讝讛讬诇 and the two newest other | |
| priority takes. | |
| ## Training | |
| The icefall `egs/librispeech/ASR/zipformer` recipe (commit `d7c83d52`): pruned RNN-T loss, ScaledAdam with Eden, base | |
| lr 0.002, 12 epochs, batch 8. It ran on one NVIDIA L40S (Nebius) on 2026-11-17. Code: | |
| `training/gefilte/` in Daniel's training repository. | |
| ## Results (held-out takes, greedy streaming decode) | |
| | | Base | Gefilte PoC-1 | | |
| |---|---|---| | |
| | WER | 21.2% | 13.5% | | |
| | CER | 7.7% | 3.0% | | |
| | 专讜讝讛讬诇 recognised | 0 of 2 | 2 of 2 | | |
| There are only 8 held-out takes, from one speaker, so treat these numbers as a sanity check, not a benchmark. | |
| Every held-out take (reference, then base, then Gefilte): | |
| | Reference | Base | Gefilte PoC-1 | | |
| |---|---|---| | |
| | 讗驻砖专 诇专砖讜诐 讗转 讝讛 注诇 砖诐 专讜讝讛讬诇? | 讗驻砖专 诇专砖讜诐 讗转 讝讛 注诇 砖诐 专讜讝讬诇? | 讗驻砖专 诇专砖讜诐 讗转 讝讛 注诇 砖诐 专讜讝讛讬诇? | | |
| | 讬砖 诇讬 转讜专 讗爪诇 讛专讜驻讗讛, 注诇 砖诐 讚谞讬讗诇 专讜讝讛讬诇. | 讬砖 诇讬 转讜讗专 讗爪诇 讛专讜驻讗讛 注诇 砖诐 讚谞讬讗诇 专讜讝讛专. | 讬砖 诇讬 转讜讗专 讗爪诇 讛专讜驻讗讛 注诇 砖诐 讚谞讬讗诇 专讜讝讛讬诇. | | |
| | 讛拽讜谞讟讬讬谞专 谞讜驻诇 讻诇 驻注诐 砖讛讜讗 注讜诇讛. | 讛拽讜谞讟讬讬谞专 讗讜 驻诇诐 讻诇 驻注诐 砖讛讜讗 注讜诇讛. | 讛拽讜谞讟讬讬谞专 谞讜驻诇 讻诇 驻注诐 砖讛讜讗 注讜诇讛. | | |
| | 诪住讚 讛谞转讜谞讬诐 讗讬讟讬 讛讬讜诐, 讗讜诇讬 讞住专 讗讬谞讚拽住. | 讞讬驻讛. 诪住注 讝讛 谞转讜谞讬诐 讗讬讟讬 讛讬讜诐, 讗讜诇讬 讛住专 讗讬谞讚拽住. | 诪住注 讝讛 谞转讜谞讬诐 讗讬转讬 讛讬讜诐, 讗讜诇讬 讛住专 讗讬谞讚拽住. | | |
| | 讛诐 讗诪专讜 诇讬 砖诇讗 讬讻讜诇讬诐 诇注砖讜转 讗转 讝讛. | 讛诐 讗诪专讜 诇讬 砖诇讗 讬讜讻诇讬诐 诇注砖讜转 讗转 讝讛. | 讛诐 讗诪专讜 诇讬 砖诇讗 讬讻讜诇讬诐 诇注砖讜转 讗转 讝讛. | | |
| | 讗转讛 诇讗 讬讻讜诇 诇讛转讞讘讗 诪诪谞讬. | 讗转讛 诇讗 讬讻讜诇 诇讛转讞讘讗 诪诪谞讬. | 讗转讛 诇讗 讬讻讜诇 诇讛转讞讘讗 诪诪谞讜. | | |
| | 讗转讛 爪专讬讱 诇砖诪讜专 注诇 注爪诪讱. | (same) | (same) | | |
| | 讗谞讬 专讜讗讛 讻诪讛 讝讛 诪住讜讻谉. | (same) | (same) | | |
| WER and CER count words only: punctuation is removed from the reference and the output alike. On the 84 training takes | |
| Gefilte scores 1.1% WER, so it has memorised them; only the held-out rows above say anything. | |
| ## Files | |
| - `model.pt`: icefall state dict under `model`, config under `config`, tokenizer under `tokenizer`, the same layout as | |
| the base. | |
| - `bpe.model`: SentencePiece tokenizer, unchanged from the base. | |
| - `sherpa-onnx/`: `encoder.int8.onnx`, `decoder.int8.onnx`, `joiner.int8.onnx`, `tokens.txt` for | |
| [sherpa-onnx](https://github.com/k2-fsa/sherpa-onnx) (`model_type="zipformer2"`). They are used exactly like the base's | |
| files; see the base card for code. | |
| ## Licence and credit | |
| CC BY 4.0, as the base. Credit: Yakov Kolani and [ivrit.ai](https://www.ivrit.ai) for the base model and its data; | |
| [Dicta](https://dicta.org.il) for the nikud model used to prepare the prompts. | |