DTMF Tone Generator & Decoder

Press a key on an old telephone and it plays a short two-note chirp. That sound, dual-tone multi-frequency signalling, or DTMF, is how a phone told the exchange which digit you had pressed. The sounds are still used by menu systems, entry-phone panels and radio equipment. This page plays them, saves them as a WAV file and works the other way: it listens to a recording or the microphone and reports the keys it hears.

How a key sounds

The keypad is a grid of four rows and four columns. Each row has a frequency in the low group, 697, 770, 852 and 941 Hz, and each column one in the high group, 1209, 1336, 1477 and 1633 Hz. Pressing a key plays its row tone and its column tone together. The digit 5 is 770 Hz plus 1336 Hz. The fourth column, 1633 Hz, gives the extra keys A, B, C and D. The frequencies were chosen so that no tone is a harmonic of another, which keeps the pairs distinct.

A worked example

Press Sample, which enters 5551234 with 100 ms tones and 60 ms gaps. Seven keys at 160 ms each last 1.12 seconds, and the WAV file at 8,000 Hz holds 8,960 frames. Press Play to hear them, or Download WAV for a file named dtmf-5551234.wav. The Sample button in the microphone section decodes that same generated sound and reports seven keys found and the digits 5551234. The keypad on the page also plays: each button sounds its tone and adds the digit to the sequence.

Settings

Tone length sets how long each key sounds, from 30 to 1,000 ms. Phone keypads typically play tones for around 50 to 100 ms. The gap between keys is the silence between tones, 20 to 1,000 ms. Keys outside the set are ignored, so you may type spaces or dashes in the field without effect. The download is 16-bit mono at 8,000 Hz, the telephone sample rate, so a file made here plays back as if it were over a phone line.

Decoding

Drop a recording of touch-tones, in WAV, MP3 or another audio format, and the page reports the digits with the time at which each started. It also offers a live microphone mode, shown as Heard live. The decoder uses a Goertzel filter bank, an efficient way to measure the energy at eight specific frequencies. A key is accepted only if exactly one tone from each group stands out and the pair lasts at least 35 ms, so a spoken word or a click does not produce digits.

Edge cases

Compressed files can blur short tones; very short beeps may be missed. Noise, music or speech over the tones can cause missing or wrong digits. Playing a sound from a speaker into the microphone adds room echo, which can merge fast sequences. Longer gaps and tones make live decoding more reliable. Decoded times are measured from the start of the file.

Limits

The page produces and reads sounds; it does not dial a number or connect to anything. It is meant for learning, testing audio equipment and checking recordings. Do not use it to try to gain access to systems you are not authorised to use. Files are processed on your device.

Related tools

• Tone Generator plays any single frequency.
• Audio Converter changes the format of the WAV you saved.
• Mic Test checks that the microphone hears the keypad.
• Speaker Test confirms the speakers reproduce the tones.