Make a mark
Type an identifier and mark a music file with it. The watermark is mixed into the music at a low level and the marked file is written to your computer. Stereo files stay stereo.
Up to 63 characters. A URL or IRI works well, because that is what a mark is for. A longer identifier needs a longer track: about 12 seconds for a few characters, 26 for 22, and about a minute for 63. Text that is not plain ASCII takes more.
The watermark is a faint noise-like signal spread through the music, built to survive MP3 compression, resampling, gain changes and cropping. It is shaped to sit under the level a model of hearing says the music hides, band by band and moment by moment. A model is not a listener, and nobody has yet confirmed by listening that it is inaudible on any given track: play the result against the original before you publish it. The tones are the original WebBeeps scheme, written as audible beeps in a new file; they replace the audio, and a music file is not used.
A phrase that decides where the watermark hides. The same phrase is needed to read it back, and anyone without it cannot find the mark. Leave it empty to use the public default key, which anyone can use, so a mark made that way can be found and read by anyone.
WAV or MP3, mono or stereo. The marked file is written as a WAV at the same bit depth as the original (an MP3 becomes 16-bit). Nothing leaves this page.
No file chosen.
Your marked file
Read a mark
Choose a marked file and read the identifier back out of it. It looks for the watermark first and then for the original tones. A file with no mark says so rather than reporting something.
The phrase the mark was made with. Empty reads marks made with the default key. A mark made with a key cannot be found without it, and the page will say "no mark found", which is also what a file with no mark says.
WAV or MP3. Nothing leaves this page.
No file chosen.
What was found
About this
This is a port and extension of WebBeeps, Danny Ayers' audible tone watermarking, which was live at webbeep.it and did the same two jobs in the same page. The tone scheme is unchanged so that old files still read. The watermark scheme is new and is what the site is for.
Everything runs in this page: the audio is decoded here, the codec is here, and nothing is sent anywhere. How it works has the detail.