Is it possible, in theory, to use sound waves as the physical medium to transmit data over a network?

In other words, could you implement layer 1 of the OSI networking model using sound waves or am I totally misunderstanding the physics/networking concepts?

  • 9
    Well, modems used to use acoustic couplers, which used sound waves to transfer data.
    – Ron Maupin
    Jan 21, 2019 at 5:50
  • 5
    Doesn't this happen every time you speak words? Jan 21, 2019 at 18:11
  • 7
    Well, you can use pigeons as the layer 1, so sound waves should be totally feasible.
    – dim
    Jan 21, 2019 at 18:29
  • 4
    As an aside, the original hack that jailbroke the original iPod/iPhone was playing the OS executable file as audio through the earphone and reverse modulating that to figure out/guess what the private key for the data encryption was. The system was locked down digitally and any attempt to transfer data (including mp3 files) only got encrypted data. But the audio subsystem (by virtue of needing to interface to the unencrypted human ear) was not encrypted (the bug of being able to play random files has since been fixed)
    – slebetman
    Jan 22, 2019 at 1:04
  • A lot of things can be used for Layer 1. Do you have a reason to think you couldn't?
    – Mast
    Jan 22, 2019 at 14:12

8 Answers 8


It's very much possible. Even excluding the old acoustically coupled modems that eventually evolved into a direct connection to a phone line, there are also programs that will let you use a sound card as a modem (I've used some before for out-of-band communications when debugging an Ethernet driver, though I used direct audio cabling instead of actual acoustic signaling), and the general concept is becoming rather popular with IoT devices for pairing with a control app on a smart phone during setup (though this is closer to an RFID tag approach).

This approach has a number of pretty significant downsides though:

  • It's very low bandwidth by modern standards. Even with ultrasonic frequencies, you're still looking at no more than a few hundred kilobits per second under good conditions. This makes it significantly less than useful except for passing very small amounts of data (such as with the IoT usage I mentioned above, where it usually just passes a 802.11 hardware address and some auth info so a Wi-Fi Direct connection can be established).
  • Outside of very limited situations, it's really slow even aside from the signal frequency. The speed of sound in air is about 340 m/s (give or take a few dozen m/s based on temperature, pressure, humidity, and air quality), which is insanely slow compared to electrical signals or electromagnetic waves (which propagate at roughly the speed of light), which means that compared to Wi-Fi or Ethernet, signal latency is rather high. This doesn't matter much for very short-range communications, but once you get past a few meters, the latency starts to get noticable (imagine if the link between your computer and your router had a longer RTT than the entire rest of the network path to this website combined). Even the best conductors of sound can only get at best 35-40 times the speed of sound in air, which is still insanely slow.
  • It's extremely sensitive to the environment. Ethernet is solid enough that it doesn't even need shielded if you have decent cabling. Wi-Fi can get bad at times, but is still at least able to reliably screen for specific frequency bands easily, and EMI is usually pretty easy to find and stop. Vibrations and sound are everywhere though. Again, this is more of an issue for longer-range communications, but it's still more of a problem than for Wi-Fi, partially because of the next point.
  • Sonic transmitters with a high transmit power are rather dangerous, both to the environment, and to people. To reliably get any kind of range beyond about a few meters before the SNR gets so high that you have no recoverable signal, you need to be operating at high enough sound pressures to cause permanent hearing loss. That much energy can also pretty easily damage delicate objects.
  • "a few meters" in the last paragraph seems like an exaggeration. I can talk to someone a few meters away without having to yell so loud that I damage my own hearing.
    – user253751
    Jan 21, 2019 at 21:36
  • @immibis, talking to someone a few meters away has a very low data rate and tremendous amounts of error correction.
    – Mark
    Jan 21, 2019 at 22:20
  • @Mark yes, but it does "reliably get any kind of range beyond about a few meters before the SNR gets so high that you have no recoverable signal" without "operating at high enough sound pressures to cause permanent hearing loss".
    – user253751
    Jan 21, 2019 at 22:45
  • @immibis You pretty much need to use ultrasonic frequencies for this type of thing to minimize interference from people talking, moving, and just existing in the near vicinity. Frequency affects how sound propagates, and higher ones need higher sound pressure at the source to reliably excite a pickup at a given distance. Try using the acoustic pairing on an IoT device from more than a few meters away, and you'll notice it doesn't work reliably, and that's a low bandwidth usage. Jan 22, 2019 at 2:29

It's been done. I used to have a 300-baud acoustic layer 1 device.


Anything that can transport information can be used as physical layer - sound waves as well as pigeons.

Some air-gap attacks use (ultra)sound to communicate across the air gap.

However, since the frequencies even for ultrasound are rather low (some kHz), the data rate would be low as well (some kbit/s). Also, the reach of sound waves limits the use to a single room.

  • About "single room". Not necessarily true, because sound waves (like almost any other waves in principle) can be amplified - take a look at SASER Jan 22, 2019 at 13:27
  • @AgniusVasiliauskas Yes, of course - but practically the reach is very limited as is the usable bandwidth. You can do a lot to improve the transmission quality but then again, why not just use RF, copper or fiber in the first place?
    – Zac67
    Jan 22, 2019 at 15:49
  • I'm not saying that we must use sound waves for information transportation. Of course nothing will beat speed of light in electromagnetic radiation. However, there may be the cases when other options fits badly - for example in strong external EM fields (solar wind, etc.), so that shielding can be impractical too for EM communication to operate. Maybe in this scenario we could try to use sound-waves-based info transportation ? (Think about having pigeons in an uninhabited island, when some ship goes across) Jan 22, 2019 at 17:10

Yes it’s possible. In fact, Amazon Dash buttons use ultrasound as a medium.


Certainly. Some options not discussed in other answers, but closer I think to the heart of your question:

  • A shotgun mic receiver and a similar transmitter. Uses air as the medium.
  • A piezo sensor receiver and a similar buzzer. Uses a rod of (say) wood, carbon fiber or beryllium rod as the medium.

In both cases, an electrical impulse provided to the transmitter will produce a mechanical impulse through the medium to be detected by the receiver, where it is converted back to an electrical impulse.

In both of these examples, we have not used any electricity to represent sound, but have actually used sound waves traveling through different media.


One of the first computers used soundwaves within a medium of mercury to store data so though it was not a networking medium it was still a method data retention and the same concept could be used for transmission and networking


I've done this before, not because it made sense to, but because I felt like it. Ham radio has all kinds of things for this, notably AX.25. Anything you send through a non-digital radio is audio being encoded into radio waves, and back by the receiver. Take the radio out of the equation, and you get what you're looking for.

Take a look at AX.25, as well as other digital modems, such as MT63 and PSK. fldigi is capable of many, although it's built for text, so you'd need to base64 any binary data.

  • AX.25 isn't transmitting using sound waves though, unless you're using acoustic coupling of the signal somewhere. It's still an EM transmission (more correctly, AX.25 is a protocol for encoding digital information on an analogue signal, no sound needed or involved). Jan 22, 2019 at 2:18
  • @Austin Hemmelgarn It's encoded in such a way as to be sent out an audio device. So while it often doesn't get sent out speakers, it makes sense to think of it as sound. Jan 22, 2019 at 2:21
  • But it's still not sound until it's actually played through speakers, a piezo element, a tesla coil, or some other device that converts the electronic signal to vibrations in the air. Jan 22, 2019 at 2:24
  • @Austin Yes. And that is possible. And I have done it. What's wrong with my answer? Jan 22, 2019 at 2:26
  • 1
    AX.25 is not a physical layer, and it's not sound waves. The signal it produces can be converted to sound waves as a physical layer, but your answer doesn't really make that clear. Jan 22, 2019 at 2:29

There was the TCP/IP over bongodrum project. But the homepage for the actual project seems dead now.

Here is a reference to the project

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.