The partnership allows Ignitarium to accelerate implementation of compact neural networks on Arm platforms, especially targeting Audio AI solutions on Arm Cortex-M class devices.Ignitarium, a global product engineering company with expertise in Video and Audio AI solutions, today announced it has joined the Arm AI Partner Program. Cost sensitive Audio applications like off-line voice recognition, noise suppression or sound effect classification need to be implemented on memory and compute-limited micro-controllers. Ignitarium’s focus on creating highly optimized neural networks targeting the Arm® Cortex®-M processor family allows this segment to be effectively addressed at the optimal power-performance-cost point.“Having worked for several years on building AI solutions targeting GPU, FPGA and other accelerators, we have developed a deep understanding of complex neural network architectures. Over the past year, we have been leveraging this pedigree to implement purpose-built tiny neural networks, enabling us to implement even video AI applications on micro-controller class devices. Our latest offering in this space is an On-Device Voice Command Recognition solution using our proprietary audio pre-processing and neural network pipeline, requiring only 38 KB RAM and 65 MHz clock speed on a Cortex-M4 based MCU. With our initiation into the Arm AI Partner Program, we look forward to greater collaboration opportunities with Arm and its wider partner ecosystem. We are especially excited about the introduction of the Arm Cortex-M55 and the Arm Ethos™-U55 microNPU and look forward to delivering advanced solutions around these in the future,” said Sujeeth Joseph, Chief Product Officer at Ignitarium.