| --- |
| license: cc-by-4.0 |
| language: |
| - en |
| - he |
| base_model: Hebrish/Sabich |
| base_model_relation: quantized |
| pipeline_tag: automatic-speech-recognition |
| library_name: mlx |
| tags: |
| - mlx |
| - parakeet |
| - code-switching |
| - hebrew |
| - english |
| - automatic-speech-recognition |
| extra_gated_prompt: "Access to this model is by request and is governed by the Hebrish terms of use (https://hebrish.app/terms). Non-commercial use is free; commercial use needs a commercial licence from access@hebrish.app. Please also complete the attestation form: https://hebrish.app/legal/hebrish-terms-of-use.pdf" |
| extra_gated_fields: |
| Name: text |
| Organisation: text |
| Intended use: text |
| "I agree to use this model under the Hebrish terms of use (non-commercial unless I hold a commercial licence)": checkbox |
| --- |
| |
| # Sabich v1.0: MLX |
|
|
|  |
|
|
| <img src="https://hebrish.app/img/sabich/sabich-512.png" alt="Sabich logo" width="140" align="right"> |
|
|
| MLX weights for Apple silicon of [Sabich v1.0](https://huggingface.co/Hebrish/Sabich), the kitchen-sink Parakeet TDT 0.6B v2 fine-tune: English-Hebrew code-switching plus tech vocabulary, brand names, numbers in digits and Irish-English phrasing, in one |
| speaker's voice. Hebrew words come out in Latin romanization. Results, training data and limitations |
| are on the main card. This build is **Sabich v1.0** (run `r7-all`); it is replaced when the line has a new version. |
|
|
| **Website: [hebrish.app](https://hebrish.app/models/sabich)**. Use of this model and every converted |
| or quantised build of it is governed by the [Hebrish terms of use](https://hebrish.app/terms). Access is by |
| request: use the button on this page, or write to [access@hebrish.app](mailto:access@hebrish.app). |
| Non-commercial use (personal, research, education, open-source projects) is free; please complete the |
| [attestation form](https://hebrish.app/legal/hebrish-terms-of-use.pdf). Commercial use needs a commercial |
| licence. Versions released under CC-BY-4.0 stay under that licence; the terms do not restrict rights it grants. |
|
|
| ## Usage |
|
|
| [parakeet-mlx](https://github.com/senstella/parakeet-mlx): |
|
|
| ```bash |
| pip install parakeet-mlx |
| parakeet-mlx audio.wav --model Hebrish/Sabich-MLX --fp32 |
| ``` |
|
|
| ```python |
| import mlx.core as mx |
| from parakeet_mlx import from_pretrained |
| model = from_pretrained("Hebrish/Sabich-MLX", dtype=mx.float32) |
| print(model.transcribe("audio.wav").text) |
| ``` |
|
|
| [mlx-audio](https://github.com/Blaizzy/mlx-audio): |
|
|
| ```bash |
| python -m mlx_audio.stt.generate --model Hebrish/Sabich-MLX --audio audio.wav --output-path out --format txt |
| ``` |
|
|
| This is not an iPhone build: the Hebrish apps, and an iOS app, use the sherpa-onnx build |
| ([Hebrish/Sabich-sherpa-onnx-int8](https://huggingface.co/Hebrish/Sabich-sherpa-onnx-int8)). |
|
|
| ## Conversion and check |
|
|
| Converted from training run `r7-all` with the remap behind the stock `mlx-community` Parakeet |
| repos ([senstella's script](https://gist.github.com/senstella/77178bb5d6ec67bf8c54705a5f490bed)), |
| fp32 weights (parakeet-mlx casts to bf16 at load unless asked for fp32; it cannot load quantized |
| weights, so there is no 8-bit build). Checked with parakeet-mlx 0.5.3 on MLX's Linux CPU backend, against the NeMo |
| PyTorch output on held-out code-switching test clips: |
|
|
| | | Identical to NeMo | WER vs NeMo | WER vs reference | |
| |---|---|---|---| |
| | NeMo (PyTorch) | | | 5.4% | |
| | MLX bf16 (default) | 9/10 | 0.5% | 5.9% | |
| | MLX fp32 | 9/10 | 0.5% | 5.9% | |
|
|
| WER here is raw (cased and punctuated) on a small sample; the main card has the full evaluation. |
| The conversion is a direct key remap (no retraining or quantization). |
|
|
| ## Base model and licence notices |
|
|
| This model is a build of [`Hebrish/Sabich`](https://huggingface.co/Hebrish/Sabich), a fine-tune of [`nvidia/parakeet-tdt-0.6b-v2`](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2), by NVIDIA, licensed under [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/). Hebrish modified the base model by fine-tuning it and converted the weights for this build. |
|
|
| The licence stated at the top of this card covers Hebrish's contribution; the base model's licence keeps applying to the base model, and nothing here restricts what that licence allows for it. Terms of use: <https://hebrish.app/terms>. |
|
|
| Hebrish is not affiliated with or endorsed by NVIDIA. |
|
|
