Evolution of Speech Recognition
Any machine that can interpret the language and speech of us humans, has been a point of fascination since the early years. In this era of voice assistants like Alexa and Siri, we never think about how speech recognition evolved through the years. In this post we will highlight how speech recognition has evolved.
Audrey – The Digit Recognizer
The first known speech recognition system could interpret only numbers. The “Audrey” system could recognize a single voice saying these digits aloud. Bell Laboratories developed this in the year 1952.
Next came the ‘Shoebox’, in the 1960s. It was developed by IBM. It was a step ahead of Audrey as it could recognize numbers as well as arithmetic commands like ‘plus’ or ‘total.’
The US Department of Defence’s ARPA (DARPA) funded a 5-year program called Speech Understanding Research in the early 1970s. This led to the creation of several systems that could recognize speech. The most successful one of them was ‘Harpy’ which could recognize more than a thousand words at once.
Hidden Markov Models
Soon came the HMM models, which used a statistical approach to leap forward in accuracy levels for speech recognition. Soon there was a set of tests that provided a benchmark for comparisons for other researches. Even the shared data sets were used to train and test individual models.
ASR was introduced in the nineties. ‘Dragon Dictate’ was launched then and could recognize approximately thirty words per minute. In the year 1997, a more evolved version was introduced named ‘Dragon NaturallySpeaking,’ which could capture words with more accuracy and pace.
Google Voice Search
With continuous improvements and research, Google managed to launch the ‘Google Voice Search’ application in the year 2008. Further refinement and a huge volume of training data led to the launch of the ‘Google Assistant,’ which is widely used even now!
Then came the era of voice assistants like Microsoft’s Cortana, Apple’s Siri, and Amazon’s Alexa. This has brought a complete transformation in the speech recognition industry. These digital assistants have brought speech recognition technology into our homes.