Is it possible to use steganography using an analog audio channel?

I know this can be done using images. My idea is to encode an audio message in way that if listened by an ordinary device (e.g., radio or telephone) it delivers a normal message. But if we use a special decoder, the receiver can listen to a hidden message. Please note that I am interested in decoding the message only using analog audio. I know it is possible if we go with digital audio.

Is this possible? If so, it would be great if someone can give me a some practical implementations of this.

  • Encode it as an analog signal below 10hz or above 40khz? Depending on the digital encoding you might be able to hide it somewhere else. Sometimes adding a second channel (stereo) but defining it as mono can help hiding info in the second channel. For images there are also several techniques for steno depending on the format. The same goes for audio.
    – Silver
    Jan 10, 2017 at 12:52
  • It is possible to use steganography on (wihtin) Audio files (digital) or in the accustic phenomenon (analog). Would you have any preference? Jan 10, 2017 at 13:14
  • most analog hardware would have trouble storing 20khz modulations in a way that could be later decoded. Microphones aren't tuned to those frequencies and most digital samplers lowpass at 20 or even 18k to reduce "noise", making it hard to transfer the watermarked audio. you might try playing with stereo channel separation; making the difference your secret audio, preferably slowed down and reversed to avoid noticeable sound.
    – dandavis
    Jan 10, 2017 at 14:18

4 Answers 4


There are a few ways to do this. Use ring modulation to add a virtual carrier to the audio you want to hide, for example at 14khz. Goldwave's mechanize feature can do this. This will give an amplitude modulation-like signal with sidebands above and below 14khz. Then use a highpass filter to get rid of the lower sideband, and only kids or dogs may hear the deedle deedle deet of the single sideband audio signal, but may not know what it is unless they're familiar with shortwave listening or ham radio. Someone else with a ringmod plugin or goldwave can apply the same 14khz effect to demodulate the signal. The normal un-hidden audio at the same time will get modulated at 14khz and will sound like hash that you can get rid of using a lowpass filter to hear the hidden audio totally in the clear, but you probably won't need to bother. I suggest before you hide your audio that you filter it with a 300hz to 3000hz filter which is what telephones and communications radios use. This will limit aliasing of the converted audio and make it harder for normal ears to hear when listening to the undemodulated file. Another method is to simply record a short message and slow it down 32 times, and mix the result with the ordinary in the clear audio you want to mask the message with. Be sure to mix it at a reasonably low volume so that the low thunder rumbles of a slowed-down human voice are not obvious under the normal audio. Then all the guy on the other end needs to do is speed up the file 32 times to hear your short message. This method needs lots of room, 32 seconds of file for each second of slowed-down audio, so make it quick.


Off the top of my head I can think of two possibilities:

  1. Hide the information in frequencies that the human ear cannot perceive (i.e. below 20 Hz or above 20 kHz)
  2. Use the spread spectrum technique to hide the information below the noise floor.

The former is not applicable to telephones since they do not transmit the frequencies which the human ear cannot perceive. This could however be used to hide data in digital audio files, there is an implementation available at [1]. The latter is applicable to phones, there is a patent available at [2]. However I am not aware of a publicly available implementation.

[1] https://github.com/solusipse/spectrology

[2] https://www.google.com/patents/US7826604


You may want to start with the following survey:

Using spread spectrum:


One well-known implementation is Backmasking, where a message can be heard when the audio is played backwards. It was a popular concept for a while, especially when records were used for listening to music because records can be manually turned backward to decode the message; although some modern audio software can also easily reverse digital audio streams.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .