Get 2 months free.

We turn ideas into brilliant solutions!

If you are looking to invest into your own speech to text recognition model, we can help. At Vatis Tech, we offer customized speech-to-text models tailored to your specific needs.


Here are some of the benefits of using customized speech-to-text models

  • Improved accuracy:by training the model on your specific industry jargon and accents, you can ensure more accurate transcriptions.
  • Increased efficiency:customized models can save time by transcribing your content faster and more accurately.
  • Flexible deployment:on cloud or on premise with your data fully secured


Customization and Key Benefits For Your Business

Higher Accuracy

Customized models can deliver between 10% to 20% better accuracy, compared to the generic models

Dedicated Support

You will receive dedicated assistance throughout the entire process.

Development Time

It will take us between 1 to 2 months to develop your model.


Easily integrate via API You only need to change the model ID.


Deployment on cloud or on premise with your data fully secured

Continuous learning

Ensures that your model stays up-to-date, even when fed with new data or any other language nuances specific to your industry

The number of hours required to train a customized speech-to-text model depends on various factors such as the quality and quantity of data, the complexity of the vocabulary, and the desired accuracy level. Whether you need to capture highly complex type of interactions, or you are working with complicated audio data (e.g. with lots of background noise, strong regional accents, etc.), a minimum of at least 10 hours of sample data, is required, in order for us to be able to put in place a custom model for you, generate labeled data and start training your model.

Generally, it can take anywhere from a few hours to several weeks to train a model, Depending on the amount of customization required, per-hour processing rates may vary, but it's worth the investment to ensure accurate transcriptions for your specific industry jargon and accents.

When it comes to deploying customized speech-to-text models, there are a tow options to consider. Either to host the model on premise or on the cloud infrastructure. Your model will be made available to you on our API-platform and can be integrated into your application in minutes with a single API call and easy-to-follow documentation.

What customers are saying about Vatis Tech speech-to-text solution


Speech-to-text applications have always been a technological challenge, given that medical language, and in particular radiological language, is a "foreign language" or rather a dialect of a strange language for most speech recognition software, requiring a lot of transcription algorithms, often with questionable results in terms of the accuracy of the transcribed text. Vatis Tech has found appropriate solutions in this context, with the accuracy of dictation transcription achieved by using artificial intelligence to process and progressively improve the results of the transcription process. It is far superior to other similar applications for the Romanian language. I am confident that this application has all the prerequisites to become in the near future one of the most efficient options for the radiologist who wants to improve his daily workflow, by dictating his reports, followed by their immediate transcription into medically, semantically and stylistically correct texts. In these circumstances, Vatis Tech promises to help reduce the time needed to write radio-imaging examination reports, which can only be very welcome from any perspective.

Dr. Răzvan Capșa
Medical Director, Primary Radiology and Medical Imaging Physician

Ready to start?The first hour is FREE!