Knight Rider is No Longer Fictional: How to Talk to Your Car with a New Vibration Microphone
ABSTRACT
Voice is the most natural way for us to interact with our surroundings. This holds true for cars as well – as they become smarter with artificial intelligence and large language models, we need easier, faster, and more intuitive ways to interact with them. Inside the cabin, this is relatively straightforward, especially as the consumer electronics and automotive worlds converge. High-performance MEMS microphones, algorithms, and DSPs used in smartphones, headphones, and smart speakers are making their way into cars. But what happens on the outside?
Automotive applications generally, and external environments specifically, pose significant challenges for MEMS microphones. These microphones require a sound port for sound pressure waves to pass through and move the membrane of the capacitor formed with the backplate, generating the sound signal. In a car, such a sound port is exposed to stressors like dust, pressurized water, and chemicals. This exposure necessitates extensive protection around the sound port, increasing costs and degrading acoustic performance.
The key to solving this challenge for external automotive MEMS microphones lies in eliminating their main weakness. In other words: No sound port, no problem.
The new Infineon Vibration Sensor (IVS) offers a unique solution. It merges the MEMS microphone technologies needed for good acoustic performance with the robust design demanded by the outside environment of a car. The IVS uses a modified MEMS microphone that makes the microphone susceptible to sounds transported through solid media, i.e., vibrations. Since it operates through physical contact rather than air movements, it does not require a sound port. The IVS can be mounted in the chassis, windshield, or other locations to pick up voices and sounds addressing the car from the outside.
Initial tests at Infineon have proven the system concept. Combined with state-of-the-art AI algorithms, the IVS enables real conversations with a car from the outside. This presentation will cover the application and use cases for external sound sensing, technical and implementation challenges of classical MEMS microphones, and the potential solution with the new Infineon Vibration Sensor and AI signal processing.
For the first part, the presentation will focus on external sound sensing for voice. We will briefly discuss new legislation, AB1777, which requires autonomous vehicles in California to be equipped with a two-way external communication system for voice.
We will then discuss why “classic” MEMS technology struggles in outside environments, both from a MEMS design and robustness perspective, as well as an acoustic one, with sound channel design and protection degrading physical performance. Infineon will discuss the vibration sensor MEMS design and explain why it helps in the aforementioned applications. The ASIC – the first one in the Infineon portfolio using a new clock frequency principle – will also be touched upon. Such a “vibration microphone” design will also be compared to other solutions like accelerometers.
Lastly, Infineon will introduce a holistic solution with both the sensor hardware and AI algorithms for voice processing. We will discuss how such AI software solutions need to work in real-time, yet be small enough to not take up too much SoC space and run on the edge. Apart from voice processing, a HW-SW solution for speech recognition commands will also be introduced.
BIOGRAPHY
Julian Kornprobst is an experienced professional specializing in MEMS microphones and their integration into cutting-edge systems, particularly for automotive and consumer applications. With a strong academic background, including master’s degrees in Chemistry and Science, Technology, and Policy from ETH Zurich, the Vienna native bridges the gap between engineering, technology, and customer needs.
Currently working as the MEMS Sensor System Competence Team business lead at Infineon in the semiconductor industry, he focuses on creating value for customers by combining technical expertise with a deep understanding of system-level applications. This approach ensures innovative and robust solutions for end-users, particularly in the realm of voice communication and new MEMS microphone technologies.