Who would have imagined that the origins of voice search date back to 1952? As voice technology becomes increasingly popular; both in homes and on the move, we thought it would be interesting to take a moment and digest where it all began.

When was voice recognition invented?

Interestingly, the first speech machine was actually invented in 1774 by Wolfgang Von Kempelen, indicating that the idea of speech machinery has been developing for over 230 years! However,true voice recognition technology didn’t really begin until the 1950’s…

Voice recognition technology timeline

The key dates in the history of voice recognition are:

history of voice recognition timeline


In 1952, the first voice recognition device was created by Bell Laboratories and they called it (her) ‘Audrey’. ‘Audrey’ was ground-breaking technology as she could recognize digits spoken by a single voice; a massive step forward in the digital world.


10 years later in 1962, IBM demonstrated a machine called ‘Shoebox’ at the World’s Fair in Seattle. Shoebox was the next huge development as it could understand 16 words in the English language.

Also, in the 1960s laboratories in the United States, Japan, England and the Soviet Union developed hardware that was dedicated to recognise spoken sounds; stretching the speech recognition technology to support four vowels and nine consonants.


Fast forward to 1970… Speech recognition takes off! The U.S Department of Defence contributed heavily towards the development of speech recognition systems. From 1971-1976, they funded the DARPA SUR which was the first official ‘speech understand’ research program.

Also in the 1970s, Carnegie Mellon University developed ‘Harpy’ which could comprehend 1,011 words and employed a more efficient system of searching for logical sentences.

The 70s continued to show huge developments in Voice Recognition technology, including the opening of the first commercial speech recognition company ‘Threshold Technology’, and Bell Laboratories introduction of a system that could interpret multiple voices. Speech recognition vocabulary jumped from a few hundred words to thousands of words, and had the potential to recognise an unlimited number of words due to the ‘Hidden Markov’ model.


Moving into the late 1980s, and voice recognition emerged for the family market! In 1987 a doll was created by Texas Instruments, known as ‘Julie’ who could be trained by children to respond to their speech.


The 90s saw the introduction of a faster microprocessors, and speech software became feasible. In 1990 the company Dragon released ‘Dragon Dictate’, this was the world’s first speech recognition software for consumers. Following ‘Dragon Dictate’ they launched ‘Dragon Naturally Speaking’ in 1997. Although you could speak 100 words a minute into the new model it would cost you a pricey $695!


By 2001 speech recognition development came to a bit of a standstill, and although computer speech recognition topped 80% accuracy by the end of the decade, work in the area seemed to be stalling.


However, in 2010 Google added ‘Personalized Recognition’ to voice search on Android phones and software could then record user voice searches to produce a more accurate speech model. The following year they added voice search to Google chrome browser and voice technology started to grow in popularity and usage. Google’s English voice system can now understand 230 billion words from user queries!

Voice search has massively developed over the decades but can we predict what lies ahead? How important is it to us as marketers to consider voice search in our marketing mix? Data suggests there were 4.2 billion digital voice assistants worldwide in 2022. And with the likes of Siri and Alexa now prevalent in people’s everyday lives and even controlling different equipment in their homes, it's to no surprise that voice assistants carry out 3.1 billion voice searches each month.

We recently hosted an exciting debate discussing the pros and cons of voice search – check out our highlights video below!

Back to blog