Reliably recognize speech

Speech Enhancer

Kardome's AI-driven signal separation and noise reduction technology facilitate a seamless voice recognition experience in any acoustic environment, from the quiet to the chaotic

Speech Enhancement technology for ASR engines
How it Works

Learn How Kardome Can Improve Your Device Voice User Interface

The critical difference between Kardome and other audio front-end solutions is that our technology clusters speech signals based on location.

Kardome’s source separation software treats each person in any environment as if they are the only person talking. This focus contrasts direction-based technology, such as beamforming, which provides limited performance indoors or in any closed environment.

Kardome’s innovative voice user interface technology enables people to give clear, understandable commands to their devices even in adverse acoustic environments. 

voice user interface performance
Core technology

VUIs and ASR Performance

An ASR’s ability to accurately translate acoustic speech signals depends on the clarity of the input signal to the ASR. As a result, noise reduction, echo cancellation, source separation, and other components are added to the VUI to enhance the acquired signal before reaching the ASR.

Kardome’s core technology includes Speech Separation, Echo Cancellation, and Noise Reduction modules that facilitate reliable ASR performance in noisy and multi-speaker scenarios.

Understand Users Anywhere, Anytime

Advanced Voice Enhancement

How often have your coworkers had trouble understanding you during a conference call?

Office environments typically contain multiple sound sources in addition to the main speaker. Sound interference makes the user’s speech unintelligible, negatively impacting important meetings and work.

Kardome’s technology enables ASR engines to understand the user anywhere, anytime. Kardome mitigates interfering signals by up to 30 𝑑𝐵 in acoustically challenging conditions that would ordinarily impede beamforming technology’s performance.

Enhanced speech recognition for conferencing