Terms of Use

These terms cover the Ri AI Assistant website and its live demo. Ri is an independent research project by Tynrai AI, based in Meghalaya, India. By using the site or the demo you accept these terms. If you do not accept them, please do not use the demo.

The Ri app has not been released. When it launches it will have its own terms.

The demo is for evaluation

The demo exists so you can judge how our Khasi speech models perform. It is not a product and is provided free of charge.

You may use what the models produce to evaluate them, including listening, reading, sharing an example with a colleague, writing about the results. You may not use the output commercially, resell it, or build a service on top of the demo. If you want to use these models in your own work, write to us.

We may change, limit, or switch off the demo at any time, and we do not promise it will be available or that any result will be reproducible later.

How you may not use it

Speech synthesis makes some kinds of misuse easy, and Khasi has very few tools for detecting synthetic audio. So these limits matter to us. You may not use Ri to:

• imitate a real person, or produce audio meant to make someone believe a real person said something they did not say;
• create anything unlawful, fraudulent, harassing, hateful, or intended to deceive, defame, or intimidate;
• upload a recording you do not have the right to share, including recordings of other people made without their knowledge or consent;
• present generated audio as a genuine human recording where doing so would mislead the people who hear it;
• make decisions about health, law, safety, employment, or anyone's rights on the strength of what the models produce;
• overload, scrape, probe, or work around the limits of the service, or use it to train a competing model.

You are responsible for what you put in and what you do with what comes out. We may block access to anyone using the demo this way.

The output will sometimes be wrong

Everything the demo returns is machine-generated. Transcripts may contain errors. Speaker labels get confused. Synthesised speech mispronounces words. This is normal for speech models, and especially so for a language with as little training data as Khasi.

Treat every result as a draft that a person needs to check. Do not rely on it where being wrong would matter.

Your audio

Audio you upload or record stays yours. You give us permission to process it for the single purpose of returning a result to you. We do not use it to train models and we do not keep it after the request finishes.

The Privacy Policy explains where it travels while that happens.

Age

You need to be 13 or older to use the demo or join the waitlist. If you are under 18, please ask a parent or guardian first.

No warranty, and what we are liable for

The site and demo are provided "as is", without any warranty of any kind. We do not guarantee that the models will be accurate, uninterrupted, error-free, or suitable for any particular purpose.

To the extent the law allows, we are not liable for loss or damage arising from your use of a free evaluation demo, including decisions taken on the strength of its output.

Our work

The name Ri, the Tynrai name, the site design, and the models themselves remain ours. Using the demo does not give you a licence to any of it beyond the evaluation described above. Our published model weights carry their own licences on Hugging Face.

Changes

If these terms change we will update the date at the top of this page. They will be replaced with fuller terms when the Ri app is released.

Contact

Questions about these terms, requests to use the models, or reports of misuse: tynrai.labs@gmail.com.