Proud to share a new milestone from EaseAudio！

Our team has developed a DNN-based Noise Suppression algorithm for embedded audio systems, with strong performance and efficient DSP loading.

Current performance highlights:

16 ms frame size / 16 kHz sampling rate

Low MIPS consumption

Only ~20% loading on a 400 MHz HiFi4 DSP

The algorithm has been successfully validated on both Cadence HiFi4 and HiFi5 platforms. To make customer integration easier, EaseAudio has also wrapped this solution into a ready-to-use Audio Weaver module.

We have already completed verification on NationalChip (GX8008C&GX8008) and Amlogic (A113X2) SoCs, demonstrating strong portability across multiple embedded platforms.

What is especially exciting is that this algorithm shows outstanding performance in babble noise and wind noise conditions, which are often among the most challenging real-world scenarios for voice products.

At EaseAudio, we continue to focus on turning advanced AI audio technology into practical, production-ready solutions for embedded devices.

If your team is looking for an efficient and proven AI-based noise suppression solution, feel free to contact us.

#EaseAudio #NoiseSuppression #DNN #AudioDSP #AudioWeaver #HiFi4 #HiFi5 #EmbeddedAI #SpeechEnhancement #WindNoiseReduction #BabbleNoise #Amlogic #NationalChip #VoiceAI