I know that the signal was just tone pulses but why was it when (back in the 90s) when you first connected to the internet you heard a bunch of funny noises. After that if you were to use the internet, it still was using the telephone line, why no funny noises then?

  • 17
    Those were the days.. – Thomas Sep 20 '12 at 12:01
  • 2
    You really needed that noise when you were paying for the phone calls, too. – Lee Kowalkowski Sep 20 '12 at 12:20
  • 7
    @Thomas: Indeed. I wonder how many kids nowadays have even heard that sound? FWIW, I miss that sound a lot (modems/routers nowadays just seem 'soulless' in comparison), although one thing I certainly don't miss is desperately trying to muffle the darn speaker with a pillow when connecting at night, before I thankfully figured out the AT command mentioned by tylerl below! (Oh, and I ran my own BBS for a while - how many people remember those? Ah, fun times.) – Karan Sep 20 '12 at 15:29
  • 1
    @JanDoggen that's not the question, I was wondering why they only did it at the beginning. – Celeritas Sep 20 '12 at 15:39
  • 6
    I miss the days when mainframes used to make a noise. A healthy ICL 2900 mainframe would chirrup like a bird. If you got a couple of seconds silence followed by by a sort of whirring sound you knew the system had crashed and was dumping its memory to disk. The next sound would be the bang-bang-bang of the line printer printing out the core dump. Happy days! – Paul Cager Sep 20 '12 at 18:10

Modems originally allowed you to send data over a network that was designed to only carry voice. Because of that, the communication method between two modems had to be in the audible hearing range (or it would not get carried on the phone line). This is no longer needed because the phone system can now carry both voice and data at the same time (DSL).

The sounds were there all the time, you just needed to pick up the phone to hear it. The reason they played it over a loudspeaker to start with is so you could hear if somthing went wrong with the connection (busy signal, wrong number, a person picked up instead of a modem on the other end, etc).

  • 71
    +1 for audible busy signal, wrong number, "hello .. hello .. damnit!" etc. – RedGrittyBrick Sep 20 '12 at 8:07
  • 6
    +1 The name 'modem' itself is very descriptive (ie modulator-demodulator). – HaydnWVN Sep 20 '12 at 9:37
  • 6
    It also happened occasionally that the modem couldn't get a carrier, and it would get stuck at repeated buzz-BONG-buzz-BONG like sounds before giving up because of a timeout (which also could be set with the appropriate AT commands, I think it was ATS7=nn actually with nn the carrier-detect timeout delay in seconds). With experience, you can fairly accurately tell from the noises whether the connection will succeed long before the modems have established a carrier, as well as the transmission speed of the resulting connection. – user Sep 20 '12 at 11:03
  • 19
    @MichaelKjörling "With experience, you can fairly accurately tell from the noises whether the connection will succeed..." Somehow this experience has never and will never make it on to my resume but man, some days I miss it. ;) – Unkwntech Sep 20 '12 at 16:55
  • 10
    Don't neglect the "At least this way I know it's actually DOING SOMETHING!" aspect -- You could usually turn off the speaker during negotiation (or turn it on full-time) with AT commands, but modem carrier negotiation was never particularly fast. Users power-cycling modems or aborting and retrying the connection because it wasn't negotiating fast enough was actually a problem for some environments. – voretaq7 Sep 20 '12 at 21:25

Because the modem speaker was turned on by default, to give the user the feedback that something was happening during the handshake. With the proper setup of the AT commands you could have 3 modes - always on for speaker, totally silent during operation, and the default with speaker turned on during connect.

They were ATL and ATM if I remember correctly.

But the whole command (Hayes commands) to the modem was:

Attention. Loudness level x.
AT      Lx   (where x is 0 to 3)

Of course this was usually part of a longer string, and many instructions were set by default (unless specifically over-ridden).

Newer modems were able to be set; and stored a default command list.

  • 3
    And if you used them long enough, you could actually tell what was going on. – Journeyman Geek Sep 21 '12 at 0:48
  • 1
    Correct; we used to debug by listening to these noises and figuring out which use cases were misbehaving. – Danny Schoemann Sep 24 '12 at 9:46

The whistles and chirps and buzzes that you hear when a modem is going through its initial handshake process is a test of the telephone line quality. A modem send precisely specified sounds and the other listens see what it actually hears on the other end. This way the modems know how clear the line is between them and what sort of frequencies they can use to communicate with each other. The more frequencies they can use and the lower the noise, the higher the speed they'll be able to communicate at.

If a connection ever failed due to connection quality, it would generally fail during this initial handshake process. And if you were listening, you could usually tell why (e.g. you got an answering machine on the other end instead of a modem).

As such, modems were usually configured to play this handshake sequence out loud. This was configured by sending AT M1 to the modem during setup. Alternately, AT M2 means to leave the speaker on all the time, while AT M0 means don't turn the speaker on at all. See the AT command set for more information.

The actual transmission noise that you would hear if you picked up the phone during an active session (as opposed to during this handshake procedure) just sounds like static.

  • 3
    At 300 baud, it's possible to audibly hear incoming data. On occasions, I've turned on the modem speaker if I wanted to hear when characters arrived on a generally-idle line. Higher baud rates use a "data-scrambler" circuit so that most patterns of data are no longer audibly distinguishable. – supercat Sep 20 '12 at 16:57
  • 6
    Back in the day (and before my voice broke), I used to be able to connect to 300 baud modems. That is, scream at the right frequency. Wikipedia's article on modems implies this was at 1,070 Hz or 1,270 Hz, if my vocal chords were the originating "modem". – ChrisInEdmonton Sep 20 '12 at 20:51
  • 5
    While once upon a time I was able to whistle at 110 baud successfully enough to get printable characters, I was never able to control it well enough to predict what characters I'd see. It was fun to try, though. Those were the days when single bits were pure tones at specified frequencies, and the Carrier Detect line meant something. – RBerteig Sep 21 '12 at 0:43

Those noises are the process of the 'hand-shake' going on between your modem and the modem of your ISP. In a literal sense, your modem is calling another modem, much like a classic telephone.

Once the connection is established (after the hand-shake is successful, i.e. the ISP's modem 'picks up'), there is no more need for any 'calling'.


The first modem I ever used was acoustically coupled--that is, you put the handset into a rubber dingus where transmitted/recieved the sound to/from a microphone/speaker on the modem body.

This was necessary for a while in the US because AT&T had government granted veto over the attachment of any electronic device to their wires---a rule that was later overturned.

On those things you could hear a little leakage if you were running it in a quite room.

  • Say, that's interesting. I never knew why the acoustic couplers went away. I assumed the technology got better. I should have known! :) – Mark Allen Sep 20 '12 at 22:16
  • Yup, only AT&T had the right to connect to the wires. Everyone else had to go acoustic. For an immense sum of money, it was possible to lease (but never buy) an audio patch device that could be used in place of the coupler. There were even electronic phone book devices that could whistle touch tones at the handset to make calling from pay phones easier. Oh, and there were pay phones too. Acoustic couplers even worked with pay phones if you were desperate. (cf. War Games.) – RBerteig Sep 21 '12 at 0:47
  • Same with BT in the UK - approved devices had a green circle on their base, and I think unapproved devices had a red triangle. At first there were no modems then there were USR modems with red triangles on them and then eventually green circles started to appear. – Rich Sep 21 '12 at 8:08

The technical reason is that because modems work over phone lines, which are by and large used by human beings to make voice calls, it behooves us, in data communication equipment going over the voice network, to have an amplifier and speaker which monitor the audio signal on the line during connection establishment.

This lets us hear things like busy signals, or the voice of a human being if we happen to dial a telephone rather than another modem.

Of course, since the signals are all mixed into one line, we cannot just hear the voice or just the busy signal without also hearing the modem's signals.

Note how, at some point in the establishment of the data connection, the monitoring is disabled. This is a feature of the modem: it squelches that amplifier and speaker because the monitoring has served its purpose, and its continuation would be an annoyance.


The name MODEM is a contraction of MOdulator-DEModulator. Modems transmit data by Modulating a signal (tone), and receive data by Demodulating the signal (tone). The sound they generate is the modulated signal. By using a tone, they can transmit a digital signal over an acoustic (sound) channel. The original modems with the cups for the headset were also known as acoustic couplers.

The original modems used a simple signalling system. As speed increased, the signalling system became more complex. Noise on the line degrades the available speed. Higher speed signalling mechanisms have error correction, and speed adjustment mechanisms built into the protocol.

North American phone systems used to charge a premium for data conditioned lines. However, any phone line which was function within specifications was capable of carrying a 1200 bit/s signal. Higher speed signals did require a much better signal, and some lines which were fine for voice use would fail to carry the signal at the full rate.

Modern digital phone systems carry sound using a digital signal. In North America, the signal is at 56 kbit/s. This is the upper limit on modem signals traversing a digital switch in North America. Last I knew, European used a 64 kbit/s channel. I don't know if Modems in Europe are/were capable of carrying a 64 kbit/s modem signal.

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