---
license: other
license_name: multiple
license_link: https://huggingface.co/Promptengineering/whryte-models/blob/main/LICENSES.md
---

# Whryte Models

Model artifacts downloaded by the [Whryte](https://whryte.com) desktop dictation app for
Windows. This repository mirrors upstream releases so the app has a stable, owner-controlled
download source. **These are not original works** — see [LICENSES.md](LICENSES.md) for the
license and origin of every file.

| Path | Model | Used for | License |
|---|---|---|---|
| `parakeet/` | NVIDIA Parakeet TDT 0.6B v3 int8 (sherpa-onnx export) | Batch dictation, file transcription | CC-BY-4.0 |
| `nemotron-en/` | NVIDIA Nemotron Speech Streaming EN 0.6B int8, 4 chunk sizes (sherpa-onnx exports) | Live dictation (English) | OpenMDW-1.1 |
| `nemotron35/` | NVIDIA Nemotron 3.5 ASR Streaming Multilingual 0.6B int8 (community ONNX export) | Live dictation (multilingual) | OpenMDW-1.1 |
| `llm/` | Qwen3-4B-Instruct-2507, Qwen2.5-1.5B/0.5B-Instruct (GGUF Q4_K_M) | Transcript enhancement | Apache-2.0 |
| `diarization/` | pyannote segmentation-3.0, 3D-Speaker ERes2Net (sherpa-onnx exports) | Speaker identification | MIT / Apache-2.0 |