Loading
Have a ComfyUI workflow? Get a readable Workflow Doc in your browser.Make a workflow readable.Try it now ↗
polyform-shield-1.0.0
Model source
Source description
An int8 ONNX conversion of NVIDIA's Nemotron speech streaming 0.6B model (the March 2026 checkpoint), packaged for on-device streaming dictation. This is the model behind Rekody, the private, local-first voice dictation tool for macOS.
Sources
1 sourceVerified Sep 13
Model artifacts
2 artifactsSource excerpts
2 excerpts
encoder.onnx (int8, 880 MB), decoder_joint.onnx (int8,
11 MB), tokenizer.model (SentencePiece), and config.json.Word error rate (WER, %), lower is better. Measured with the Open ASR
Leaderboard methodology (hf-audio/open-asr-leaderboard @ b6bdcd0b,
EnglishTextNormalizer), on identical audio and with the identical
normalizer for both models.
| Dataset | rekody-streaming-en-0.6b-int8 | openai/whisper-large-v3-turbo |
|---|---|---|
| AMI | 12.36 | 13.86 |
| Earnings22 | 12.27 | 10.81 |
| Gigaspeech | 8.64 | 8.34 |
| LibriSpeech test-clean | 2.09 | 1.60 |
| LibriSpeech test-other | 4.92 | 3.75 |
| SPGISpeech | 2.99 | 2.78 |
| Voxpopuli | 2.74 | 5.31 |
| Average | 6.57 | 6.64 |
Note the operating points: this model streams, producing text as audio arrives in 560 ms chunks, while whisper-large-v3-turbo is a batch model that sees each full recording before emitting anything. Matching a strong batch model's average WER under a streaming constraint is the point of this model.
Two licenses apply to this repository, and users must comply with both:
LICENSE). PolyForm Shield is a noncompete license: any use is
permitted except providing a product that competes with the software or
with the licensor's products built on it.NOTICE and the dated copy in licenses/).| File | SHA-256 |
|---|---|
encoder.onnx | 83538392358e90c40592f7cf99ee65ac7dca5d144edb999ce028c372318a5753 |
decoder_joint.onnx | 89cae615623e41e94cc6b428708cd1a89a22965606fe7ded814b1d8e20c87368 |
tokenizer.model | 07d4e5a63840a53ab2d4d106d2874768143fb3fbdd47938b3910d2da05bfb0a9 |
config.json | 19c6a445ddf0268a8ab3653f0cdf5740196271ba0a902eaae8d17de52f74a829 |
These hashes are also published in SHA256SUMS.
df1f0fe9dfdf05152936192b4c8c7653d53bf557 (the March 13, 2026
checkpoint).quantize_dynamic QInt8 over MatMul and LSTM ops.Contact: [email protected]
--- license: other license_name: polyform-shield-1.0.0 license_link: https://polyformproject.org/licenses/shield/1.0.0 base_model: nvidia/nemotron-speech-streaming-en-0.6b language: - en pipeline_tag: automatic-speech-recognition tags: - automatic-speech-recognition - streaming - onnx - int8 --- # rekody-streaming-en-0.6b-int8 An int8 ONNX conversion of NVIDIA's Nemotron speech streaming 0.6B model (the March 2026 checkpoint), packaged for on-device streaming dictation. This is the model behind [Rekody](https://rekody.com), the private, local-first voice dictation tool for macOS. - Cache-aware streaming ASR: transcribes while you speak, in 560 ms chunks, with encoder cache carried across chunks. No audio leaves the machine. - Runs on the ONNX Runtime CPU execution provider. The files load directly with [onnxruntime](https://onnxruntime.ai) or [parakeet-rs](https://crates.io/crates/parakeet-rs). - Four files: `encoder.onnx` (int8, 880 MB), `decoder_joint.onnx` (int8, 11 MB), `tokenizer.model` (SentencePiece), and `config.json`. ## Why trust this conversion - The conversion pipeline was validated bit for bit against the prior public int8 conversion of this architecture: when fed identical inputs (the same upstream checkpoint revision), our pipeline reproduces that artifact's tensors exactly. What ships here is the same pipeline applied to NVIDIA's current (March 2026) checkpoint. - The cost of int8 quantization was measured as statistically zero. On LibriSpeech test-clean (n = 2620), int8 vs the fp32 export it came from differs by +0.072 pp WER with a paired bootstrap 95% CI of [-0.004, +0.150], which includes zero. ## Benchmarks Word error rate (WER, %), lower is better. Measured with the Open ASR Leaderboard methodology (`hf-audio/open-asr-leaderboard` @ `b6bdcd0b`, `EnglishTextNormalizer`), on identical audio and with the identical normalizer for both models. | Dataset | rekody-streaming-en-0.6b-int8 | openai/whisper-large-v3-turbo | |---|---|---| | AMI | 12.36 | 13.86 | | Earnings22 | 12.27 | 10.81 | | Gigaspeech | 8.64 | 8.34 | | LibriSpeech test-clean | 2.09 | 1.60 | | LibriSpeech test-other | 4.92 | 3.75 | | SPGISpeech | 2.99 | 2.78 | | Voxpopuli | 2.74 | 5.31 | | **Average** | **6.57** | **6.64** | Note the operating points: this model streams, producing text as audio arrives in 560 ms chunks, while whisper-large-v3-turbo is a ba...
Source context: 6 downloads · 0 likes · Pipeline automatic-speech-recognition · Repo Rekody/rekody-streaming-en-0.6b-int8