The growing interest into controlling devices by means of voice commands has led to a massive spread of sytems equipped with Voice User Interfaces (VUIs). We find them in smartphones, home appliances, automation systems and cars.

VUIs need a clean speech signal to correctly understand the commands. However, in real scenarios, recordings are affected by different noise sources, such as babble noise due to other people talking, heating, ventilating and air conditioning (HVAC) noise, environmental noise and, in the automotive scenario, the car engine noise.

BdSound’s technologies suite (S2C) includes several algorithms to deal with noise, such as beamforming and AI-based noise reduction. Now, with the continuous improvements we apply to our technology suite, it is possible to enhance the experience of human to machine interaction.

In this video, we present our brand new technology that allows to separate and recognize voice commands coming from two different speakers talking simultaneously using only two microphones.

This solution runs on a wide variety of microprocessors, starting from just two microphones. It can be applied for voice commands from simultaneous talkers or in extreme noisy conditions.

If you want to know more about how we can assist you with your product, contact us.