Definition
DTMF, or dual-tone multi-frequency signaling, is the way a phone reports a key press. Each key plays two tones at once, one low and one high, and the receiving system decodes the pair into a digit.
Read the full guide: What is DTMF? Tones, frequencies, and voice AI
How DTMF works
Each key plays two tones at once, one from a low group and one from a high group, and the receiver decodes the pair into a digit. ITU-T Q.23 defines which pair belongs to which key. The what is DTMF guide has the frequency table.
On a call, the digit can reach the other side in two ways. In-band, the tones are ordinary audio inside the call, which a compressed codec can distort. Out-of-band, the digit is sent as an event, and RFC 4733 defines how to put such events into RTP packets so the digit does not depend on the tones surviving the codec.
Why it matters for voice agents
Speech is a poor channel for some inputs. A caller reading a 16-digit account number aloud invites recognition errors, while keypad digits arrive exactly. An agent can ask for the keypad when accuracy matters and accept speech otherwise.
Reading digits works too. On Twilio calls and the agent WebSocket, a Cartesia agent reads a caller’s keypad entry as input and replies to it. Test incoming keypresses on your own carrier path before you rely on them.
The direction can also reverse. An agent that calls a company with a phone menu has to press keys to reach a department. Cartesia Managed Agents have a send_dtmf system tool for that.
Digits typed on a keypad can still land in recordings and transcripts, so treat them as sensitive. For the full mechanics, read the guide.