Licences & Attribution
Parayu integrates open-source machine learning models, speech recognition binaries, and high-performance C/C++ inference runtimes. Every upstream copyright, model licence, and attribution requirement is listed below — this page mirrors the same Licences & Acknowledgements screen shipped inside the app.
OpenAI Whisper & whisper.cpp
Offline Speech Recognition Engine (ASR)
Powers Parayu's offline speech transcription models (LOW, HIGH, PRO, SWIFT, TURBO, MALAYALAM). Uses OpenAI's Whisper model architecture and Georgi Gerganov's high-performance whisper.cpp C/C++ inference runtime with Apple Metal GPU acceleration.
Sarvam AI (Indian AI Models)
Malayalam & Indian Language Processing
Powers the Indian AI Brain in Parayu for Malayalam, Hindi, Tamil, Telugu, Kannada, Bengali, Gujarati, Marathi, Punjabi, Sanskrit and the other scheduled Indian languages. sarvam-translate is a fine-tune of Google's Gemma 3 4B, published by Sarvam AI under the GNU General Public License v3.0.
Modification notice: the weights shipped with Parayu are quantised by RenderLogic Media & Technologies for on-device inference. The quantised file is a modified version, not a verbatim copy.
Written offer of source: RenderLogic will provide the complete corresponding source for this modified model, and the GPL-3.0 licence text, to any third party on request — write to support@renderlogic.in. Valid for three years from the date you received this copy.
Because the model derives from Gemma 3, Google's Gemma Terms of Use also apply upstream.
Google Gemma 3
International AI Brain & Let's Talk chat
Gemma 3 4B powers Parayu's International AI Brain — offline text cleanup, grammar restoration, tone adjustment and translation for European, East Asian, Middle Eastern and African languages — and the offline Let's Talk chat.
Modification notice: the weights shipped with Parayu are quantised by RenderLogic Media & Technologies for on-device inference.
llama.cpp Runtime Engine
Hardware Accelerated GGUF LLM Server
Runs both AI models above locally as an on-device inference server — no cloud calls, no API keys, nothing installed outside the Parayu app bundle.
Electron, Node.js & Chromium
Desktop Application Runtime & Framework
The desktop application shell Parayu is built on for macOS and Windows.
Questions about a licence, or need the source offer above fulfilled? Write to support@renderlogic.in.