Speaker Separation

Automatic speaker recognition (ASR) is a very difficult problem when used in noisy environments. If two or more people are talking at the same time it is getting even more difficult. ASR enginges have a hard time following a conversation if there is overlapping speech and noise. It is nearly impossible to transcribe speech properly, if there are two people talking at the same time. Smart assistants like Alexa or Siri are not working properly anymore. Nevertheless, Evolve is able to solve this ill-posed problem! Our algorithm is able to locate a specific speaker in noisy environments and cancels out unwanted background noise. For demonstration purpose, we prepared two audio recordings while driving a car. The first one contains the original recorded sound. There are two people talking at the same time and the radio is turned on.

The second recording contains the same data, but was processed with our powerful speaker separation software. The aim of this pre-processing step was to focus only on the driver talking, removing all unwanted interfeering speech and backgrounnd music. Do you hear the difference? The voice of the other person as well as the radio music is magically removed. This enables crystal clear telephone conversations and high quality automatic speech recogntion in car environments, offices or even at home, leading to a unique listening experience, you don't want to miss anymore.

Please drop us an email if you want to integrate our system into your business process.