Audio watermarking
100%
of samples marked at generation
A signal inside synthesised speech that listeners cannot hear and software can detect: a statement that the audio is synthetic, carried in the audio itself.
Why the waveform, not the file
A cloning system good enough to be useful is good enough to be misused. File headers are gone after the first transcode, so metadata is the wrong place for the marker. A watermark rides in the waveform itself.
What a watermark is not
- Not audible, it does not change what a listener hears.
- Not speaker identification: it marks audio as synthetic, it does not say whose voice was cloned.
- Not optional here, it is applied at generation on every request, part of synthesis rather than a post-process a configuration can skip.
Every term on this page is measurable. Take a key and read the numbers off your own requests.
Get a key and measure it yourself