Phased Array Microphone (2023)
A DIY phased‑array microphone with nearly 200 low‑cost capsules showcases how beamforming and delay‑and‑sum processing can “steer” listening direction, isolate sound sources, and even refocus on different locations after recording—similar in spirit to light‑field cameras. Commenters explore applications from bat tracking, leak detection, videoconferencing, and medical ultrasound to acoustic cameras and assistive tech for the hard of hearing, while also debating hardware design trade‑offs (PDM vs I²S mics, PCB layout, FPGA vs SoC) and the physical limits of spatial audio capture.
Overall Reaction
- Strong enthusiasm for the phased-array mic demo, especially the ability to “refocus” direction after recording, analogous to light-field cameras.
- Several people express a desire to play with or replicate the project; some wish similar tech were in consumer devices.
Directional Audio & Source Separation
- Users imagine editing focus post‑capture to isolate sources by location, e.g.:
- Listening to different conversations in VR as you move around a virtual room.
- Selecting one region of a room (e.g., a couch) while suppressing TV and remote participants.
- Delay‑and‑sum beamforming is highlighted as the basic technique; more advanced processing is implied.
- Some note that echo cancellation and room reflections limit how “clean” spatial isolation can be.
Real‑World and Industrial Applications
- Existing/analogous uses:
- Acoustic cameras for leak detection in compressed air systems and gas/corona discharge.
- Drone detection and aircraft deconfliction in low visibility.
- Boeing and others using spherical arrays to locate noise sources in aircraft.
- Wildlife research: multilateration and beamforming to track bats, study impact of wind turbines, and map bat “sonar beams.”
- People wonder about practical tasks like finding squealing capacitors or car squeaks.
Conference, Consumer, and Accessibility Uses
- Noted that beamforming is already standard in high‑end conferencing gear (e.g., ceiling/array mics, “Meeting Owl,” smart speakers, Kinect).
- Desired future: cooperative arrays across phones/laptops, better far‑field ASR, and live transcription with source separation for hard‑of‑hearing users.
Hardware Design & Array Geometry
- Discussion of long “arm” PCBs vs many small boards:
- Long boards reduce wiring/debug overhead; fabrication is surprisingly cheap.
- Manufacturing issues (defective boards, DFM) still significant.
- PDM mics favored over I2S/TDM for cost, pin count, and very high effective sample rates; I2S chaining is attractive but limited by available I2S ports.
- Radial pattern chosen for:
- Simple mechanical construction (hub + repeating arms).
- Good distribution of pairwise distances; grid centers are underutilized and math is harder.
- FPGA/Zynq boards and alternative FPGAs are proposed as accessible platforms.
Physics, Calibration, and “Thermometer” Effect
- Calibrating mic positions and optimizing speed of sound effectively turns the array into an over‑engineered thermometer.
- Thread digresses into:
- Speed of sound dependence on temperature, humidity, and (weakly) pressure.
- The maxim “all sensors are temperature sensors; some also measure other things,” with many examples (IMUs, batteries, soldering irons, speakers).
Privacy, Security, and Exotic Sensing
- Acoustic cameras could retrospectively “zoom in” on conversations in public spaces; seen as both powerful and creepy.
- Speculation on combining phased arrays with laser microphones or high‑speed imaging to extract audio and even keystrokes via surface vibrations; effectiveness and geometry constraints are debated.