---
license: cc-by-4.0
base_model: nvidia/parakeet-tdt-0.6b-v2
library_name: voixful
tags:
  - automatic-speech-recognition
  - apple-core-ai
  - on-device
  - palette4
  - voixful
  - privovoice
---

# voixful-parakeet-tdt-0.6b-v2

Apple **Core AI** (Palette4-quantized) conversion of [`nvidia/parakeet-tdt-0.6b-v2`](https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2) for fully on-device speech-to-text on Apple Silicon — used by the [Voixful](https://github.com/AustinJiangH/voixful) engine and the [PrivoVoice](https://github.com/AustinJiangH/PrivoVoice) dictation app. This `.aimodel` directory is downloaded by the app and loaded by Voixful's Parakeet TDT backend.

## License & attribution

Weights derive from `nvidia/parakeet-tdt-0.6b-v2` (**CC-BY-4.0**, © NVIDIA) and are redistributed under CC-BY-4.0 with attribution. The Voixful conversion pipeline is Apache-2.0.