Downloads · 30 days
0
Sunno
desigrit/Sunno
Sunno is a automatic speech recognition model from desigrit. Use it when you need speech turned into text. It is set up for onnxruntime-genai. The card lists the license as mit.
These are reproducible ONNX Runtime GenAI conversions used by Sunno, an offline live-captioning app.
Access
Public
Updated Aug 6, 2026
Repo size
339 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.data339 MB · 98%
At a glance
- Task
- Automatic Speech Recognition
- Library
- onnxruntime-genai
- License
- mit
- Access
- Public
- Created
- Aug 6, 2026
- Updated
- Aug 6, 2026
- SHA
- f57a056c
Base models
Model card
From the Hugging Face model README
Sunno Whisper models
These are reproducible ONNX Runtime GenAI conversions used by Sunno, an offline live-captioning app.
| Directory | Upstream model | Precision | Intended runtime |
|---|---|---|---|
base/ | openai/whisper-base | int8 | ONNX Runtime GenAI on CPU |
tiny/ | openai/whisper-tiny | int8 | ONNX Runtime GenAI on CPU |
The files were generated directly from OpenAI's MIT-licensed Whisper weights with
bench/convert_whisper_genai.py in the Sunno repository. They are not copies of a
third-party ONNX conversion.
Each directory is a complete model and contains genai_config.json, the audio
processor and tokenizer configuration, and the encoder and decoder ONNX graphs with
their external data files.
Reproduce
python -m venv .venv-convert
.venv-convert\Scripts\python.exe -m pip install onnxruntime-genai onnx onnx_ir transformers torch
.venv-convert\Scripts\python.exe bench\convert_whisper_genai.py --model base tiny --out dist\onnx
The conversion targets the CPU execution provider because that is the native path on Windows ARM64. Sunno performs all inference locally after the initial model download.
- Task
- Automatic Speech Recognition
- Library
- onnxruntime-genai
- License
- mit
- Created
- Aug 6, 2026
- Updated
- Aug 6, 2026