Compatible Microphone Array Development Kit for Microsoft Azure

In 2020, the Azure team set out to create a rapid development platform for edge AI applications and wanted to incorporate our audio solution into their AIoT development kit. Their main requirement was to integrate a low-latency, reliable microphone array that could work with the Azure platform’s API, so developers could easily handle audio input without needing to write or interact with embedded code.

This project presented two main challenges: first, the microphone array needed to process audio on low-power devices, while meeting the demanding real-time performance needs of edge computing; second, the solution had to work seamlessly with the Azure API, allowing developers to use the module without dealing directly with embedded system code.

Solution

To meet these needs, we customized our microphone array solution, optimizing both hardware and software for Azure API compatibility.

We adapted the existing microphone array to ensure it fit well with Azure’s developer kit, including modifications to the USB protocol stack to ensure stable, efficient data transfer between the board and the host computer. With these changes to the USB communication layer, developers were able to control and debug the solution without having to modify or work with low-level embedded code.

Throughout the process, we worked closely with the Azure team, refining hardware design through multiple iterations to confirm that each version operated reliably and delivered the performance expected. This collaboration allowed us to achieve seamless API integration and simplify the overall development workflow for the Azure team.

Results

The final solution fully met Azure’s requirements. Our improvements to the USB protocol stack made it much easier for developers to interact with the development kit, without worrying about low-level audio processing details. This enabled them to focus on their own applications rather than spending time troubleshooting the audio module.

Performance-wise, careful tuning and optimization ensured the system ran reliably in different scenarios and integrated smoothly into the Azure platform. The solution delivered the low latency and stability required for edge AI applications.