Crystal Clear Communication

The Neural Beamformer (NBF) extends your portfolio with beyond-state-of-the-art noise cancellation and selective hearing powered by artificial intelligence (AI). The system removes, isotropic and directed and moving noise sources, including overlapping speech in car interiors with high efficiency. In particular, a source to distortion ratio (SDR) up to 18dB is achieved using car series adaptation. Radio-, car- engine, road-noise-sources, and overlapping speech are removed with high efficiency. This enables crystal clear communication using VoIP systems and mobiles but also improves automatic speech recognition up to 90%.

Call Efficiency

Evolve provides lightweight, fast, memory-optimized inference code for easy product integration. The software package supports both, armhf and x64/86 architectures. Further, an easy-to-use toolchain and framework to train and adapt the NBF to acoustic scenarios are provided. A GPU-based training framework is offered to adapt the AI to car-interior specific parameters and deploy it to the target device. This reduces development costs and effort dramatically. Evolve provides technical support, including audio expert reviews and feedback to help you stand out from the market.

VoIP Communication

Solving the cocktail party problem is of great importance when using smart assistants. We can solve this problem, extracting a single person's voice while driving a car. The technology can be combined with acoustic echo cancellation and VoIP communication stacks and can be retro-fitted to your car series. By doing so we add a value proposition to your customers, helping you to stand out from the market. This leads to happy clients and happy operators, boosting your revenue streams and efficiency. Have a look at our product teasers to understand how we can help you to integrate this fascinating technology into your software solution.


AI Noise Cancellation



AI Software Modules

Quick Facts

AI powered
Beamforming
     
  • Central deployment to workstation or cloud
  • No additional servers needed
  • Up to 18 dB SNR improvement
  • Integrateable into existing software platform
  • Trained >10.000 hours of speech / noise
  • Use with all headsets or mobiles
  • Volume price discounts
AI powered Echo
Cancellation

     
  • Central deployment to workstation or cloud
  • No additional servers needed
  • Up to 18 dB SNR improvement
  • Real-time software architecture
  • Integrateable into existing software platform
  • Trained >10.000 hours of speech / noise
  • Use with all headsets or mobiles
  • Volume price discounts

AI Audio Demonstration

In the following we provide some audio examples, recorded in an car. You will hear multiple persons talking. The conversation is pretty hard to follow, due to radio and noise sources. You can click on the play button to listen to the audio examples provided. Using the switch on the side you can enable / disable our audio filter, removing unwanted background interference extracting the driver's voice in real-time enabling crystal clear communication.


Car Conversation Car Radio Car City Car Motorway Car Double-Talk
Off
On