science & techWellcome Collection1928–present
Synthetic speech began with a 1928 Bell Labs device and became the voice of screen readers
Synthetic speech began in 1928, when Bell Labs engineer Homer Dudley built the Vocoder, which broke speech into signals and rebuilt it electronically. His later Voder, shaped like a piano and played with a wrist bar and pedal, was the first electronic speech synthesiser that could produce connected speech, though it sounded electrical rather than human. In the 1960s and 1970s, reading machines paired optical character recognition with speech synthesis. The Kurzweil Reading Machine, released in 1976, cost between $30,000 and $50,000. Screen readers such as JAWS, created by blind programmer Ted Henter, made computers usable for blind people, and today's voice assistants use neural networks to add natural inflection.
The voices in phones, smart speakers and screen readers have a long history, and many of their features were driven by disabled users. The technology behind assistants like Siri and Alexa grew from accessibility needs.
The Voder was played like an instrument, using a wrist bar to produce vowel sounds and a pedal to control pitch.
Wellcome Collection · CC BY 4.0
Illustrative photo · Internet Archive Book Images, No restrictions, via Wikimedia Commons
