Overview

This project is an attempt to make echolocation more accessible by using a device that virtually slows down the sound.

The main component of the project is an application that can be deployed on a Raspberry Pi 3 and can compute the echo response of the environment using the data from a depth vision camera. The echo is similar as it would be heard if the camera would be a speaker emitting beeps, the camera had ears on both sides and the sound would travell 100x slower.

The slowdown makes the phenomena easier to comprehend.

Here is an example of listening to the echo of a tree in front of the camera. The camera moves from seeing the tree on the righ side of the frame, through a position where the tree is on the left side of the frame back to the starting side. The color markers in the image represent the part of the image that generate the echo sound that is being played at that moment. The beep with a different frequency at the beginning of each image signals the time when the virtual speaker in the camera would emit a sound. You will need stereo headphones.

One can observe that when the camera is placed with the tree on its right, the right sound channel gets the echo reply quicker than the left.

You can find more examples here.

You can also play an interactive game utilizing the same technique to render 3D space into sound here.

The author currently does not know to what extent the objects could be recognized from their echos and how usable such a device might be to visually impaired persons. In case you ran some tests, would like to or are interested in the project you can find my contact email in the github commit logs, protected with captcha here or use this one.