Aaware Far Field Voice Development Kit

Documentation

The Aaware far field voice development platform provides a complete single-chip voice and sound capture solution focused on applications where strong interfering sources and high ambient noise are present. The platform consists of a 13-microphone array shield and a custom Avnet MiniZed 7010 processing board based on the Xilinx Zynq-7000 All Programmable SoC.

The Aaware sound capture algorithms include source localization, proprietary spatial/spectral/temporal source separation, and noise suppression to deliver low distortion audio of acoustic sources in the sound field. Up to 16 microphone arrays are supported, including non-uniform 1D and 2D arrays in end-fire or broad-side configurations. 1D and 2D source localization is supported depending on array choice. The Aaware sound capture algorithms are accelerated in FPGA hardware, optimizing performance, software footprint, cost, and power.

The AEV13MZ platform also integrates Sensory’ Inc. multi-wake word for robust voice assistant applications in the presence of strong direct or ambient noise with low distortion, allowing for better integration with third party speech and natural language technologies. The Aaware technology is always on, captures the speaker as they move, and cancels multiple noise sources without needing the reference audio, making it a very flexible fit for any product that anticipates strong interfering noise sources.

AEV13-MZ Features:

Complete and Easy to Integrate Solution

- - Includes Complete Linux (Ubuntu) Development Environment
  - ALSA based audio API
  - Single Chip Xilinx Zynq Processing Solution
  - oday’s Multi-Chip Solutions = DSP + SoC

Robustness

- - Removes Interference Without Reference
  - Strong Ambient Noise Rejection
  - Multi-channel AEC Support
  - Localize Weaker Source Within Stronger Interference
  - Source Localization Passed to Downstream Applications

Scalability