Today’s focus on the history of voice over for Voice Assistants rounds off our series of Voice Over Origins! But we’re ending on perhaps the most contemporary of all the industries we’ve covered so far - though that is not to say that it doesn’t have an interesting history that has led us to where we are today. These days, we have become accustomed to having our own virtual assistant with us at all times - on our phones and in our homes. In fact, we’ve not only got used to it - many of us have become reliant on it! In a surprisingly short space of time, the voices of smart speakers have quite literally become household names.
The history of voice assistants begins with the dawn of speech recognition. Of course, prior to voice assistants actually having voices that used voice over artists, first came technology that focused on allowing for products that could be voice activated. To give a brief summary of that:
1922 - the first voice activated product was released. This first product was a toy dog called Radio Rex which appeared to wake up in it’s dog house and jump out upon hearing its name, “Rex” called out. This was done by an electromagnet that picked up on the frequency of the vowel created when the user says “Rex”. Understandably given the times, this is regarded as a somewhat crude form of voice recognition though impressive none-the-less.
I would venture to say that Radio Rex is all the more an impressive achievement in the arena of voice recognition considering that the next product to have a similar ability did not appear until 3 decades later!
1952 - At the World Fair, Bell Labs introduced the world to Audrey. Audrey was a machine capable of understanding digits 0-9 and could be used to dial phone numbers hands-free. It lacked appeal for various reasons - it took time for the machine to adapt to each user before being able to somewhat accurately capture their speech and required the speaker to pause between saying each number, meaning it was far quicker and easier to just dial manually. It was also a 6-foot tall machine that was costly to make, maintain and power.
1962 - IBM announced the IBM Shoebox at the World Fair. Similar to Audrey in that it understood digits 0-9 but also recognising six commands, plus, minus, total, subtotal, false and off. The purpose of the Shoebox was to be a voice activated calculator and was the size of, of course, a shoebox!
1976 - Harpy was launched by Carnegie Mellon University, having received funding for a speech recognition project from Defence Advanced Research Projects Agency (DARPA) in 1971. This was a huge leap for products with speech recognition. Harpy could recognise 1,011 words and was capable of understanding entire phrases. In response to not computing what the user had said, Harpy responded with a synthesised voice “I don’t know what you said. Please repeat.”
By the 1990’s, the general public had better access to their own technology in the form of computers at home and mobile phones. With this came further advancements in voice recognition, such as NaturallySpeaking in 1997 by Dragon. This software could understand human speech without the need to pause between each word and could accurately transcribe it into a document. Compared to earlier speech recognition products, this was far more affordable - but at $695, it still cost a pretty penny!
The introduction of using an actual voice over artist for voice assistants appears to be Siri! Which may not seem like much of a shocker to you. But given all the people I thought were ‘firsts’ in the different industries covered in this Voice Over Origins series, and being proven wrong time and again, I’m surprised that this really was the origin of voice over for voice assistants!
Siri was originally released in 2010 and had been recorded with an American, British and Australian accents several years earlier. The American voice over artist who originally provided the voice of Siri was Susan Bennett. The original British voice was Jon Briggs. And the original Australian voice was Karen Jacobsen.
Following Siri’s success, there have been multiple smart voice assistants, though the application of voice has been approached differently by different companies. For example, IBM’s Watson, which came out a year later in 2011, is both a computer generated voice but also comes from a specific voice over artist, Jeff Woodman. Whereas Amazon’s Alexa, released in 2014, is computer generated and not from any real person.
The leaps that were made from decade to decade have been quite staggering. As is the fact that these voice assistants are an accepted and often relied upon part of everyday life for many people - which is something that has only happened in the last decade! In terms of voice over, the idea of TTS (text-to-speech) is a concern for many. I take solace in the fact that over the years and across the industries, a desire for authenticity and conveying a sense of human connection has been the goal and the option best received by consumers. Real voice over is unique and beautiful - just like the artists themselves.