Services

...

Speech/non-speech detection

Determines speech segments in the input audio stream.

Voice to text (V2T)

Converts detected speech segments into text with time stamps for indexing.

Text post-processing

Formats the stream of recognized words into a proper form using a set of grammars.


Add your own custom words with every request.

Processing more than hours of audio every day

Supported languages for V2T


Supporting languages spoken by more than 1.3 billion of people

Slavic languages


Bosnian

Bulgarian

Croatia

Czech

Macedonian

Montenegrin

Polish

Russian

Serbian

Slovak

Slovenian

Ukrainian


Other languages


Hungarian

Norwegian

Romanian

Spanish

Swedish

US English


Accuracy varies depending on task and language. Contact us to try V2T on your own data.

Features



Streaming

All our APIs process audio and return results in real-time, so you can react immediately


Multilingual

V2T is available for 18 languages with optional live updates, custom language models and ability to add words to dictionary.


Secure

Supporting on premise installations without internet access at any scale.

Integration



REST

Batch processing of offline recordings.


Websocket

Real-time communication inside web browser.


GRPC

Non-web applications with fast response time requirements.

See our API documentation here.

Deployment



Container


Single machine


High available cluster


Cloud


Hosted

Founders


Our team is formed by computer science researchers, developers and business people with more than two decades of experience.

...
Petr Herian

Business

...
Petr Červa

Research

...
Jan Nouza

Research

...
Ondřej Smola

Platform

...
Petr Pazour

Integrations

...
Jindřich Žďánský

Platform