July 2026/Fullstack/fun project
LSBeat
This project started from a casual conversation with my lecturer about his old thesis on audio steganography. Hearing his stories sparked my curiosity, so I decided to build my own version just to see how it works under the hood. The initial prototype was strictly file-based, requiring raw .wav files for both input and output. But who actually keeps .wav files around for music nowadays? Realizing the friction, I rebuilt the application to accept something much simpler. Now, the input and output are as easy as pasting a YouTube link.
The core mechanism relies on Least Significant Bit (LSB) steganography. Audio samples are just bytes. If you change the very last bit of an audio byte, the average human ear cannot hear the difference. The change is microscopic. But that single bit is enough to hide a secret binary message. You slice your text into bits, swap them into the audio stream, and the song sounds exactly the same. To extract it, you simply read those last bits back. It is simple, elegant, and completely invisible.
To be honest, this implementation would need a complete architectural rethink if it were meant for serious cybersecurity, corporate espionage, or actual covert operations. The current encoding process is relatively basic. But for having fun and hiding secret messages in songs to share with friends or a partner, this version is more than enough.