The Speak & Spell's sound chip uses a compressed audio format called "linear predictive coding". This command will read random bytes and attempt to decompress them as if it were audio data compressed in this format, then play it. This results in a unique sound which is similar to a glitching Speak & Spell.

This is a very hackish way to do it that I'm mainly just posting for fun, and I guess technically can more accurately be said to result in undefined behavior. What the command does is tell the shell to treat libpng like it's a shell plugin (which it's most certainly not) and attempt to install a "png_create_read" command from the library. It looks for the struct with the information about the command; since it's always the command name followed by "_struct", it'll look for a symbol called "png_create_read_struct". And it finds it, since this is the name of one of libpng's functions. But bash has no way to tell it's a function instead of a struct, so it goes ahead and parses the function's code as if it was command metadata. Inevitably, bash will attempt to dereference an invalid pointer or whatever, resulting in a segfault.

Replace $CMDLINE_FILENAME with the name of the cmdline file you copied from /proc/pid, and $COMMAND with the command to execute with those arguments.

Exactly what the summary says. This command will connect to another computer via SSH, and stream audio from your computer's microphone to the other computer's speakers. Use it like an intercom.

I seem to do this compulsively every time I change directories, sometimes even when I don't, even if I know exactly what I need to do. (Don't worry, the sample output is just an exaggeration. :) Show Sample Output

Remove everything after the second "rev" to only display the duplicate filenames and their hashes.