Visual Microphone

Thu, 02 Jul 2026 00:00:00 +0000

This was my final-year project at the Middle East Technical University Physics Department, and it was probably the first time my interests in optics, acoustics, and signal processing fully collided in a useful way.

The question sounded slightly ridiculous, which is usually a good sign:

Can we recover sound without using a normal microphone?

Instead of detecting pressure directly, we looked for sound in light. More specifically, we investigated whether acoustic vibrations could be recovered from changes in the speckle pattern produced by a multimode optical fiber.

At the time, I was fascinated by the idea that a messy optical pattern could contain hidden information about the environment. A speckle image looks random, but it is not meaningless. It is a fragile interference pattern, and fragility is often just sensitivity wearing dramatic clothing.

The idea

A multimode fiber supports many optical modes. These modes interfere at the output and create a speckle pattern. If the fiber is disturbed by sound or vibration, the optical path lengths of the modes change slightly. This changes the speckle pattern.

So the logic was:

sound perturbs the fiber,
the fiber perturbs the optical speckle,
the camera records the speckle fluctuations,
signal processing tries to recover the original sound.

In less polite terms: we asked whether a chaotic-looking optical blob could be bullied into becoming a microphone.

What we built

The setup used a 1550 nm laser, a multimode optical fiber, and a camera to record temporal changes in the speckle pattern. We played sound near the system and extracted signals from the recorded optical fluctuations.

The project involved:

optical alignment,
recording speckle patterns,
extracting temporal intensity variations,
filtering and signal processing,
comparing the recovered signal with the sound played through speakers.

We even tested it with actual music, including the Inspector Gadget theme, because apparently scientific seriousness has limits.

What came out of it

The main result was that sound could indeed be reconstructed from optical speckle fluctuations.

The reconstruction was not magically clean from the beginning. It required filtering and processing, and the signal was sensitive to the experimental conditions. But that was also the point: the speckle pattern was carrying acoustic information, even if it was doing so in the most unnecessarily dramatic way possible.

Listen

Original

Reconstructed

This project became important for me because it showed that optical systems can act as indirect acoustic sensors. That idea never really left me.

Why it mattered for me

Looking back, this project was one of the roots of my current research direction.

My PhD now deals with all-optical photoacoustic imaging, where ultrasound is detected optically rather than with a conventional piezoelectric detector. The physics and hardware are different, but the taste is similar:

use light to listen to sound.

The Visual Microphone project gave me an early intuition for optical acoustic sensing, speckle-based measurement, and the fact that the useful signal is often hidden inside something that initially looks like noise.

Acknowledgements

This project was carried out during my undergraduate studies at METU. I am grateful to Berk N. Gün for his contribution to the project and to Prof. Emre Yüce for his guidance.

Sound Reconstruction via Optical Multi-Mode Fiber

Sun, 18 Feb 2024 00:00:00 +0000

This preprint is connected to the project.

PDF arXiv

Signal Processing | Ege Küçükkömürcü