56 lines (35 loc) · 1.58 KB

IMAGE SOUND MAPPER

Credits

Heavily inspired by the algorithm used in 2FA made by Samip Regmi, Raju Bhetwal, and Nayan Nembang

IMAGE TO SOUND

Following is the algorithm I used to convert image to sound:

1. RESIZE

Resizes the source image to specified image size

2. GRAYSCALE

Converts the resized image to grayscale whose pixel value can be mapped from 0-255

3. PIXEL TO FREQUENCY

Each pixel from 0-255 is mapped to frequency from 200-1000 using linear mapping

4. AUDIO FILE IS SAVED

Those mapped frequencies are saved in each 0.01 second of the audio file, holding 441 samples each
with total frequency data being size of resized image
default: 50 x 50 = 2500

SOUND TO IMAGE

1. AUDIO CHUNK DETECTION

As duration per frequency is known, we extract all the chunks of data.
Each chunk holds 441 samples, in total 2500 chunks.

2. CHUNKS TO FREQUENCY

We then use librosa to find the frequency of each chunk

3. FREQUENCY TO PIXEL

Using linear mapping we convert the frequencies back to pixels

4. PIXEL TO IMAGE

Finally, all the pixel data is saved back into the image

2025-09-22.23-27-28.mp4

CLONING

clone the repo using git clone <remote-url> u can use ssh or https
after cloning install required dependencies using pip install -r requirements.txt
add and correct the required paths in main.py
run the program using python3 main.py