Task

To perform core R&D activities and develop new DSP and machine learning algorithms embedded in Cochlear implants.

Challanges

• Algorithms must be efficient due to the limited memory and processing power available on cochlear implant devices.
• Algorithms must be resilient to a wide variety of acoustic environments and noises, such as wind, microphone rubbing, and background chatter.
• Algorithms must pass clinical studies and are subject to a high level of scrutiny due to medical device regulations.

How we helped

• Carried out core R&D activities over several years
• Researched and created several signal processing algorithms: de-reverberation, speech enhancement, microphone diagnostics, and many others.
• Contributed to various machine learning projects, including a speech-music-noise classifier and own voice detection.
• Helped create labeled datasets.
• Modified various algorithms to reduce complexity and enable fixed-point DSP porting.
• Provided documentation creation, project management, presentations, and regular liaison with the client.

Results

• Our R&D contributions helped Cochlear advance the technology of their implants and, hopefully, improve the lives of 600k users worldwide.