🔊 Welcome to the 1969 COFFEE Speech Synthesizer! ✕

Long before microchips or digital audio, this 1969 Philips NatLab machine synthesized a human voice from scratch using only electronic buzz, static hiss, and analog filters. Want to see—and hear—how four circuits build the word "coffee"?

?

COFFEE

!
1969 Philips NatLab Exhibit: Synthesizes "coffee" from 4 analog phoneme circuits (C–O–FF–EE).

As you've heard, this machine says "coffee". It does so by breaking the word down into its basic speech sounds or "phonemes". All words are composed of phonemes—there are only about 40 of them in the English language. "Coffee" is made up of four of them, as shown below.

Since each phoneme is just a set of frequencies with a certain sequence, it can be made up electronically and fed into a speaker. Theoretically, if you had a machine with the 40 phonemes and you punched the buttons in the proper order, you could say anything you wanted.

Questions, accents, emphasis, etc., involve a frequency change over the whole word or sentence, which can be done electronically by adjusting the controls below. (By the way, since this unit was built in Holland, the machine says "coffee" with a definite Dutch accent.)

AUTO-RUN: OFF
TURN
TURN
EXCLAMATION! QUESTION? BASS TREBLE
LEFT KNOB: INFLECTION QUESTION? (+75%)
◄ EXCLAMATION! STATEMENT QUESTION? ►
RIGHT KNOB: BASS / TREBLE 130 Hz (MID)
◄ BASS (85 Hz) 160 Hz TREBLE (240 Hz) ►
SEQUENCER STATUS: READY — TAP AUTO-RUN OR SPEAK ONCE
1: C (/k/) 2: O (/ɒ/) 3: FF (/f/) 4: EE (/iː/)