FAQ
Frequently asked questions
Short, factual answers. If yours is missing, write to info@graey.ge.
- What is Graey?
- Graey is a software company from Tbilisi, Georgia, building streaming Georgian speech-to-text (STT) for telephone audio.
- Is the product available now?
- No. It is in development and the model is in training. You can join the early access waitlist.
- Does it work with 8 kHz telephone audio?
- That is exactly what it is built for. The model is fine-tuned on narrowband 8 kHz telephone speech.
- Is it real-time?
- Yes, it is designed for streaming: partial transcripts while the speaker talks, then final transcripts.
- Can it run on-premises?
- Yes. It is designed to run on your own servers or in your own cloud account.
- Does our audio leave our network?
- No. In an on-premises or private cloud deployment, audio and transcripts stay in your environment and are not sent to Graey.
- What data do you train on?
- Telephone speech we have the right to use, collected with the consent of the people recorded. We do not use customer audio for training without written agreement.
- Do you have accuracy (WER) numbers?
- Not yet. We will publish word error rate on real Georgian telephone audio, with a description of the test set, once we have measured it.
- Which languages do you support?
- Georgian. That is our only focus.
- How much will it cost?
- Pricing is not set yet. Tell us about your use case through the early access form.
- How do I join early access?
- Fill in the early access form or email info@graey.ge.