Espressif Systems (688018.SH) is proud to announce that its AI Lab has developed Audio Front-End (AFE) algorithms that have been qualified as a “Software Audio Front-End Solution” for Amazon Alexa Built-in devices. This is the result of Espressif’s focus not just on the connectivity of its AIoT solutions, but also on high-performance machine learning on the edge. Espressif’s AFE algorithms take advantage of the newly announced ESP32-S3 SoC, which has integrated AI and DSP acceleration. The combination of ESP32-S3 with the AFE algorithms provides a high-performance solution for Alexa Built-in devices. Espressif’s AFE algorithms will also work on our future SoCs that will have AI and DSP acceleration support.

The proliferation of the voice-user interface (VUI), as a natural way of communication with smart devices, has consequently increased customers’ expectations of VUI performance. Thus, a VUI nowadays needs to be responsive in noisy environments or even when a VUI-enabled device is playing audio over a speech command. Espressif’s AI Lab has tackled these requirements by developing a suite of algorithms that include multi-channel acoustic echo cancellation, blind source separation (beam-forming), voice activity detection and noise reduction. These algorithms work with two microphones separated by as small a distance as 2 cm. These algorithms provide a filtered audio signal to the voice-user interface, which can then process it effectively either offline or online.

Espressif’s ESP32-S3 SoC is equipped with AI and DSP acceleration, and the AFE algorithms are optimized for use on ESP32-S3. These algorithms consume as little as 12-20% of the CPU, as well as 220 KB of internal and 240 KB of external memory. This leaves additional headroom for other applications to run on the same SoC.

Espressif provides assistance to its customers with the acoustic design of their product, so that they can focus on the core value they want to offer. For further information on the Espressif’s AFE algorithms, please contact Espressif’s customer support team who will be happy to help you as soon as possible.