Can You Recover Sound From Images?

Can You Recover Sound From Images?

by Veritasium

Formal & Physical Sciences Physics PHPhysicsPHJOptical physics
🎙 Veritasium 👥 21.1M 📅 March 1, 2019 ⏱ 11 min 👁 4.0M 📄 science communication 🧭 2026-08-27
Available in: English (current) Français

Keywords

visual microphonevibrationhigh-speed cameraimage processingsound recovery

Summary

In this video, Veritasium explores the possibility of reconstructing sound from video images alone. The host, Derek Muller, visits Dr. Abe Davis, a researcher who developed an algorithm to extract audio from high-speed footage by analyzing tiny vibrations in objects. The video explains the fundamental challenge: sound vibrations move objects only about a micrometer, much smaller than a pixel, making direct motion detection impossible. Instead, the algorithm looks at edges in the image, where subtle brightness changes correlate with movement, and sums these signals to estimate displacement over time. The initial experiment uses a crumpled foil and a standard camera at 180 fps, recovering only a basic rhythm. To improve, they use a camera at 1000 fps, and after processing, they successfully recover the tune ‘Shave and a haircut, two bits’. The video also discusses the limitations of frame rate, the need for sampling at least twice the frequency of the sound, and the potential for spying, as demonstrated by recovering speech from outside soundproof glass. The sponsor segment about LastPass is integrated at the end, highlighting the security implications of sound recovery from typing.

185 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides a clear and engaging explanation of a complex scientific concept. It demonstrates the process through hands-on experiments, showing both the challenges and the successes. The argumentation is solid, grounded in the research of Dr. Abe Davis, and the video does not oversimplify the technical details. The step-by-step approach, from low-frame-rate to high-frame-rate, effectively illustrates the importance of sampling rate. The inclusion of a real researcher adds credibility and depth.

Scientific Rigor, Source Quality, Title Accuracy

The video is based on peer-reviewed research by Dr. Abe Davis and colleagues, and features direct collaboration with the researcher. The methodology is explained clearly and the limitations are acknowledged. The sponsor segment is clearly separated and does not affect the scientific content. The title accurately reflects the content, and the video does not make exaggerated claims. The description provides links to Dr. Davis’s website and the music source, but no direct links to the research papers are given.

166 words

Title / Content Match

The title accurately reflects the content: the video demonstrates and explains the process of recovering sound from images, including the scientific principles and practical challenges.

Quality & Reliability

8/10

The video is based on peer-reviewed research by Dr. Abe Davis and colleagues, and features direct collaboration with the researcher. The methodology is explained clearly and the limitations are acknowledged. The sponsor segment is clearly separated and does not affect the scientific content.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

  • Abe Davis - Personal Website — Website of Dr. Abe Davis, the researcher featured in the video, providing access to his publications and research.
  • Epidemic Sound — Source of the background music used in the video.

Concurring Sources

Contribution & Novelties

The video provides an accessible demonstration of a cutting-edge research technique, making the concept of a ‘visual microphone’ tangible for a general audience. It highlights the practical challenges and the importance of high-speed imaging. The collaboration with the original researcher adds authenticity and depth.

Pour aller plus loin :

91 words

Radar Profile

The radar profile shows high scores in information quality and reliability, reflecting the solid scientific basis and clear presentation. The technical level is moderate, making it accessible to a broad audience while still conveying the complexity of the method. The quantity of information is good, covering both the theory and the practical demonstration.

Reliability 8/10

💬 Très positif. Sur les 30 commentaires analysés, le public est enthousiaste et amusé par la démonstration, saluant la rigueur scientifique et l'humour de l'expérience.