The idea
Phones around a room align their microphones to locate sounds and explore different balances from a band practice or gathering.
How it works
- Several phones record from different positions and estimate their clock offsets.
- The microphone array uses timing and spatial differences to estimate sound direction.
- A listener can select a voice or instrument and explore its recorded balance.
AI’s role: Label likely sound sources and help isolate a requested voice; acoustic timing, beamforming, and separation need signal processing and measurement.
First demonstration
Place four phones around a musician, record a sharp hand clap at the start, align the recordings, and compare the estimated sound direction from each position.
What to solve next
How much clock error and room reverberation can the array tolerate before source direction becomes unreliable?