ProductUpdated on 13 February 2025

ReconVox: Dual-engine Speech Recognition Robust to Noise

Rafael Delgado

CTO at DTec Biometrics

Madrid, Spain

About

ReconVox is our highly efficient Speech Recognition engine robust to noise, capable of high accuracy rates under difficult acoustic conditions.

Thanks to our exclusive AutoLearn feature, the recognition engine can adapt on the fly to challenging conditions in the audio signal in order to keep top accuracy rates on the output transcription. You can adapt to a specific speaker, a noisy environment, a low quality audio channel or even a strong local pronunciation.

The recognition engine can also be configured with custom language models and vocabularies defined from scratch by the user for maximum flexibility and accuracy.

In addition to the free speech natural language mode, ReconVox supports rule-based, strict syntaxes and a Word-Spotting mode for detecting keywords in real time under extreme conditions: very noisy acoustic environments, unstructured speech where language models are useless, multiples languages mixed up, etc.

ReconVox is available as a SDK developed in C++ for maximum efficiency, speed and easy of integration.

Looking for

  • License partner

Applies to

  • Integrated processes and IT solutions (Digital Factory)

Organisation

DTec Biometrics

Company

Madrid, Spain

Similar opportunities