When looking for a way of testing hearing aid algorithms outside of the lab, a team of researchers realised a Raspberry Pi could be a sound investment. So now Raspberry Pi boards are being used in hearing aid research.
For millions of people around the world, hearing aids are hugely beneficial. Not only do they allow people to hear better, they have been shown to lower the risk of dementia, the potential for loneliness, and the likelihood of people withdrawing from social situations.
This article first appeared in The MagPi 67 and was written by David Crookes. Click here to download your free copy of The MagPi 67.
But while a lot of research has gone into developing hearing aids over the years, Tobias Herzke, a signal processing engineer at HörTech in Oldenburg, Germany, says: “There is still a lot of potential for improvement, especially in acoustically difficult situations.”
Next-gen medical research tool
For that reason, the company – a spin-off of the University of Oldenburg – developed openMHA, which was designed to be a common, portable software platform for hearing aid research and teaching. “The openMHA platform allows for real-time audio signal processing with low delay,” explains Hendrik Kayser, who develops signal-processing algorithms for digital hearing devices.
By providing a set of standard algorithms to form a complete hearing aid, openMHA can process the signal from a live microphone and perform different tasks such as amplification, directional filtering, noise reduction, and feedback suppression. Testing new algorithms is not always straightforward, however, which is where the Raspberry Pi can come into its own.
Hearing Aid Research: Quick Facts
- openMHA can process audio signals in real-time
- It can be adapted to an individual’s hearing loss
- Delays between input and output audio is below 10 ms
- There’s no GUI – except for fitting the amp parts
- Funding came from the US National Institutes of Health
Hearing Aid Research: Testing times
“The openMHA software can execute on Linux computers in a laboratory environment, but the sound environment in a lab will always differ from real sound environments encountered by hearing aid users in real life,” Tobias tells us. In the past, this has often led to skewed results that do not offer a true reflection of how hearing aids are used. Yet using ARM-based devices such as the Raspberry Pi offers a wonderful solution.
By running openMHA on a Raspberry Pi and taking advantage of its portable nature, the researchers can evaluate new algorithms in real-time and in realistic outdoor conditions. “It has allowed us to implement a new algorithm on a mobile device in order to find out how it sounds in real-time when running around with a hearing aid,” says Tobias. As such, it gives research in this field a real boost by providing better quality and more relevant data to work with.
Hearing loud and clear
It certainly beats lugging a Linux laptop around and it has also enabled the researchers to cut costs. “Small systems like the Pi provide decent computing capabilities at lower power consumption and small size,” adds Hendrik, who says the Raspberry Pi has the best price-to-performance ratio and loves the bonus of on-board wireless. “They have allowed us to offer a complete, working hearing aid signal-processing chain and keep the cost below $300.”
The knock-on effect of this has been significant. “By using the Pi, we have lowered the entry barrier for hearing aid algorithm development and evaluation, for students at university and amateur hearing enthusiasts who lack the resources to invest in large-scale setups,” says Marc René Schädler, who is leading a hearing search group at the university. What’s more, by using openMHA for the software, Raspberry Pis, and commercially available headsets and sound cards, the team is able to ensure the system is stable and flexible.
“Low cost, high availability through off-the-shelf components, mobility, no proprietary solutions, community support, and easy access to hearing aid software were the most important driving points here,” concludes Marc.
Step-01 Grab a Raspberry Pi
Originally, portable testing of the openMHA software platform was carried on on a Linux laptop. Marc René Schädler, leading a hearing research group at University of Oldenburg, felt a Raspberry Pi would work better.
Step-02 Add a sound card
Marc took a Raspberry Pi 3, loaded the Raspbian Stretch OS on to a 32GB microSD card, snapped up an Audio Injector Stereo sound card, and cloned the openMHA source code.
Step-03 Calibrate and set up
According to Marc, calibrating the setup before using it is vital. “Hearing aid algorithms need to know the absolute physical sound of the input,” he says. This has been worked into the project.