AI-powered, automated, and super-accurate speech recognition infrastructure tool for all of your audio and video data.
Made with ❤️ and ☕ in Bucharest
That’s how accurate our competitors are, on average. And they’re also 4x more expensive.
Vatis is more accurate even without any training, and it gets even more accurate over time.
We transcribe the audio data with the current version of our Speech-To-Text model. We split the result into fragments that can be easily analyzed, corrected, and validated. Also, we start an initial self-supervised training process for our technology at this step.
Our team of validators starts to analyse, correct and validate the data from the previous step. They take unlabelled data and label it.
When we have enough new hours validated by our team, we re-train the Speech-To-Text model using a supervised technique this time.
When the training has finished, we deploy the new version of the model. We are also constantly researching better ways to improve our model's architecture.
Vatis is continuously learning. We repeat the steps above until we push our accuracy beyond human — it usually takes around 1-2 months to get to that level.
Executive Director, JURIDICE.ro