◈ Steganography Suite

Hide It
In Plain Sound.

Three tools for hiding data inside audio and video. Mix cipher tracks into carrier audio, encode images into spectrograms, or embed secret audio into video files. Everything runs locally — nothing leaves your browser.

Audio-in-Audio Steganography

Mix an encoded cipher WAV into any carrier audio — music, ambience, speech, or a silent track. The cipher hides inside the sound at a low volume, inaudible to casual listeners but readable by FFT analysis or the AudioCipher decoder.

Cipher Audio (WAV from encoder)
⊙
Drop cipher WAV
or click to browse · any audio format
Carrier Audio (music / ambience)
♫
Drop carrier audio
or click to browse · WAV, MP3, OGG, FLAC
Cipher volume mix   15%
Carrier volume   100%
Cipher start offset   0.0 s
Output length

How low should the mix be?

5–20% is the sweet spot: the cipher is inaudible under music but fully readable by FFT analysis. Higher values make it more audible but harder to miss when scanning.

What carriers work best?

Ambient music, white noise, rain recordings, or simple instrumental tracks. Busy tracks with lots of high-frequency content can mask the cipher; simpler carrier = easier decode.

Decode the mixed file

Load the mixed WAV into the AudioCipher decoder, set the correct cipher mode, and run the FFT analysis. The cipher will be recovered even under moderate carrier volume.

Image-in-Audio Spectral Hiding

Encode an image into a WAV audio file so that the image appears when the audio is run through a spectrogram analyser. The technique — made famous by Aphex Twin's hidden face in "Windowlicker" — uses the audio's frequency content to draw the image. Optionally mix into a carrier track to hide it further.

Source Image
▓
Drop image here
or click to browse · PNG, JPG, WEBP, GIF
Carrier Audio (optional — for hiding)
♫
Drop carrier audio (optional)
or click to browse · leave empty for pure image-audio
Frequency range
Min Hz
Max Hz
Duration   6 s
▾ Advanced settings
Image amplitude  0.70
Spectrogram preview
▓ Drop an image to preview spectrogram output

The Aphex Twin trick

Richard D. James (Aphex Twin) encoded a spiral self-portrait into "Windowlicker" (1999), visible only in a spectrogram. The image was discovered by fans years later using spectrum analyser software.

Best images to use

High-contrast images with clear outlines — logos, text, simple portraits, QR codes. The higher the frequency range, the more vertical detail you can fit. Keep images under 300px wide for clean results.

Plausible deniability

Use the optional carrier mix to hide the image-audio inside music. The resulting file sounds like normal audio but reveals the image under spectrogram analysis — a classic ARG discovery mechanic.

Audio-in-Video Embedding

Embed a hidden cipher audio track inside a video file. Choose to replace the video's original audio track, mix the cipher in underneath it, or add the cipher as an ultra-quiet hidden layer. Output is a WebM file with the modified audio — fully in-browser, nothing uploaded.

Video File
◈
Drop video here
or click to browse · MP4, WebM, MOV, MKV
Cipher Audio
⊙
Drop cipher audio
or click to browse · WAV, MP3, OGG
Embed mode
Cipher volume  10%
Original audio volume  100%
Cipher start offset  0.0 s
Loop cipher
Recording video 0%
The video plays at 1× speed to capture all frames. Do not close this tab.

How recording works

The video plays at 1× speed in a hidden canvas element. Frames are captured and combined with the modified audio using the browser's MediaRecorder API. No server involved.

Mix vs. Replace vs. Ultra-quiet

Mix: cipher sits beneath original audio (10–15% is inaudible). Replace: original audio removed, cipher only. Ultra-quiet: 1–3% cipher buried deep — for forensic-level detection only.

Output format

Output is always WebM (VP8 + Opus), the only format browsers can encode natively. Most video players and editors accept WebM. Use ffmpeg to convert to MP4 if needed.

File size

WebM output is typically 2–4× smaller than the original MP4 at similar quality. Recording happens in real-time so it takes as long as the video duration.