By Industry

By Technology


Microphone array meets AI-based approach

With the help of AI, Using AI for Edge Speech Enhancement. Utilise AI to perform more accurate data analysis.


Microphone array meets AI-based Approach

The two microphones work together to make up for the voice signal's amplitude attenuation caused by its forward movement, particularly for high-frequency components, clear tones, and other phenomena. An array is used to create a spatially focused beam that increases gain in a particular direction while removing interference noise coming from the opposite direction.



Challenges with Using AI for Edge Speech Enhancement.

Calculations for AI algorithms are typically quite complex and involve a lot of matrix multiplication. Maximising the effectiveness of the supported operators is a crucial factor to take into account while designing the architecture.

How to quantize and prune the model to fit the neural network model in the restricted RAM while maintaining the noise reduction performance and speech quality of the algorithm.

It is significantly more difficult to deploy an AI voice augmentation solution in runtime on a Bluetooth SoC than it is to do it offline. A significant amount of commitment is needed, along with close coordination between the algorithm vendor and chip maker, for the system integration.


The Edge AI Voice Enhancement by Arm and SoundPlus.

As long as the Arm IP and instruction set are employed, it can be swiftly installed even if the hardware is changed. This facilitates a faster speed to market and a reduction in total development time and effort.

On the MCU, a real-time AI voice improvement model is running. In this step, models are created and tuned, the inference is optimised, data is compressed (int8), and data streams are managed.

True wireless stereo headphones, including those with up to four microphones, have adopted an AI solution. With outstanding S-MOS scores in the objective test, the AI speech-enhancement model successfully enhances the voice quality of true wireless stereo headphones in contexts with dynamic interference.



Technology will contribute to the definition of chipset requirements for wireless headphones

Therefore, the ARM MCU on truly wireless headphones makes it possible for dynamic noise reduction and high-quality speech enhancement to be performed in real-time at a sampling rate of 16KHz.

Artificial Intelligent Role in Data Security services

The industry needs artificial intelligence now more than ever because the world is smarter and more connected than ever. According to many reports, cyberattacks will become more persistent over time, necessitating the use of AI tools by security teams to maintain the integrity of systems and data.

Speak with Experts

Interested in implementing AI in your business with AIACME?

Take the first step by contacting us!