Speed, Accuracy, Scalability

VRX Features



Verbyx VRX is a ground up design suited to most applications of speech recognition

VRX has no arbitrary use limitations (e.g. language model size restrictions)

VRX uses existing standards wherever possible

Speed, Accuracy, Scalability

Key Features

Speech recognition is a sophisticated technology with an almost infinite number of configurations. VRX has many features that make it suitable for almost any application of speech-to-text.

The Verbyx team are also proficient in identifying and developing new features to assist in making your projects a success.

Ask us for a comprehensive description of VRX capabilities and our custom development services.

Live Mode

Online-mode optimized for minimal latency and has ability to receive live partial recognition results


Permits the inclusion of new words (not in the dictionary) in a dynamic grammar at run-time without explicitly providing a pronunciation.


Batch Mode

Offline-mode optimized for maximum throughput

Speaking Rate

Automatic and dynamic real-time adjustment to address the issue of speaking rate variations for an individual speaker or groups of speakers

Noise Suppression

Dynamic adjustment for environmental noise conditions


Supports a range of applications, from single-channel desktop to multi-channel server configurations


Flexible deployment including cloud, on-premis and B2C & B2B applications

Integration & Deployment

VRX is designed to be flexible and scalable in it’s deployment. However, the Verbyx team are unable to foresee all possible use cases and configurations. 

We are excited to discuss your specific technical and business needs with you.

Ask about custom API and operating platform development


Native API utilizes C (C89 compliant) to maximize integration flexibility and binary stability


C++ API provides a safer, more modern, interface for C++ applications

Language Model

W3C SRGS (ABNF and XML) standards compliant

Dynamic grammar support
Support for extra-large grammars

ARPA n-gram statistical language models are supported

Semantic Interpretation

W3C SISR (Semantic Interpretation for Speech Recognition) standard, fully supported

Semantic results available in XML or JSON format

Multiple interpretations can be provided for a single recognition result in grammars with semantic ambiguity (allowing application to select the valid interpretation for the situation)

Scroll to Top