Speech recognition and voice assistants
Machines that recognise spoken words, evolving from digit recognisers into conversational voice assistants on billions of devices.
- 1952

First version
Bell Labs (K. H. Davis, R. Biddulph, S. Balashek) — Audrey
Image: Bell Telephone Laboratories (Bell Laboratories Record photo) · © Bell Telephone Laboratories / Nokia Bell Labs; no licence stated · Bell Laboratories Record, vol. 31, February 1953, p. 52 (PDF page index 9), 'Automatic Digit Recognizer'; scan by worldradiohistory.com - 1962

IBM Shoebox
Recognises 16 spoken words including digits; shown at Seattle World's Fair.
Image: IBM Archives · © IBM Corporation (IBM Archives photo; … · IBM history site - 1976

Harpy (Carnegie Mellon)
DARPA project recognises about 1,000 words using graph search.
Image: Ed Thelen · Unclear; no licence stated on ed-thelen… · Ed Thelen's computer history pages (photo of a … - 1997

Dragon NaturallySpeaking
First consumer continuous-dictation software.
Image: See below · Public domain · Wikimedia Commons - 2011

Siri (iPhone 4S)
Voice assistant ships on a mass-market phone.
Image: Apple · Public domain · Wikimedia Commons - 2012

Deep neural network speech recognition
Deep networks replace Gaussian-mixture acoustic models at Toronto, Microsoft, Google and IBM, cutting errors, sometimes by wide margins.
Image: Microsoft Corporation — Jason Wells · Public domain · Wikimedia Commons - 2014

Amazon Alexa / Echo
Far-field voice assistant in a home speaker; Google Assistant follows 2016.
Image: Frmorrison at English Wikipedia · CC BY-SA 3.0 · Wikimedia Commons - 2025

Alexa+ (generative AI assistant)
Assistant rebuilt on large language models holds natural conversations and completes multi-step tasks online.
Image: Amazon · Public domain · Wikimedia Commons
Contested: Siri grew from SRI's DARPA-funded CALO project; attribution between SRI, Siri Inc. and Apple is often blurred.