June 2, 2026
Speak easy? More like freak easy
SaySynth: A Brief History of Speaking Machines
From creepy dolls to robot voices, readers are obsessed with the weirdest talking machines
TLDR: SaySynth uses Apple’s built-in computer voice system, but the bigger story is its tour through centuries of eerie talking machines. Readers zeroed in on the uncanny side fast, adding even creepier examples and turning the chat into a mix of awe, laughs, and mild existential dread.
A maker built SaySynth, a music-style tool using the voice system already tucked inside Apple computers, then took everyone on a wildly haunted tour through the history of machines that try to talk like humans. The article jumps from an 1700s hand-operated speech contraption to a 1939 machine played by trained women called Voderettes, to those deeply unsettling Edison dolls and the retro computer voice S.A.M. that helped shape later Mac speech. In plain English: this is a love letter to how fake voices went from spooky mechanical mouths to the software voices we casually hear today.
But the real popcorn moment came in the community reaction. One commenter instantly tossed in Hideyuki Sawada’s research as an even more uncanny chapter in the saga, basically saying: oh, you think this history was creepy already? Buckle up. That one link completely steers the mood of the discussion toward the internet’s favorite flavor of tech commentary: equal parts admiration and "absolutely not sleeping tonight." It also backs up one of the strongest feelings bubbling under the piece — that the history of talking machines is not just clever, it’s fundamentally weird, theatrical, and a little cursed.
There’s also a quieter hot take buried in the article that readers are likely to latch onto: the people operating these machines, especially women, did the hard part and got sidelined while inventors took the glory. So yes, this is a story about speech tech — but in the comments, it’s already becoming a story about creepy robot culture, forgotten performers, and how fast fascination turns into nightmare fuel.
Key Points
- •The article introduces SaySynth as a synthesizer built on macOS’s text-to-speech framework and frames it within a longer history of speech synthesis.
- •It classifies speaking machines into four types: mechanical, formant/rule-based, sample-based (concatenative), and generative neural/AI systems.
- •Historical examples covered include von Kempelen’s 1773 speaking machine, Joseph Faber’s Euphonia, Edison Talking Dolls, and the 1939 VODER.
- •MUSA, developed in Italy in 1978, is described as one of the first practical diphone synthesizers using about 2,000 phoneme-transition recordings plus DSP smoothing.
- •S.A.M. from 1982 is identified as the first commercially available speech synthesizer and as a precursor to Macintosh’s Macintalk technology.