Page 162 - Grasp English B1+ (Student Book)

|
Science Mechatronics

We can see the beginnings of speech
recognition technology in 1952 when an
American company developed Audrey, an
automatic digit recogniser. At the time, Audrey
was a wondrous machine. It could listen to a list
of numbers (0–9) and flash the corresponding
light, demonstrating its ability to recognise the
word. Audrey was an enormous machine and,
unfortunately, was usually only accurate when
working with trained speakers.

This highlighted a significant issue with speech
recognition. Each person has a different voice
and factors such as individual pronunciation
needed to be taken into account. A. Discuss with a partner.

1. Does anyone in your family have a voice-
However, developers were determined to
improve on the work that Audrey’s team had activated assistant, such as Amazon’s Alexa or
started. In 1962, IBM showed the world their Apple’s Siri?
ShoeBox, a machine which could understand 2. What things can voice-activated assistants
sixteen words in English. Then, ten years later, currently do?
Harpy was created. It was able to understand 3. What do you imagine they’ll be able to do in
the same amount of vocabulary as an average the future?
three-year-old, at just over one thousand words.

These machines, and those which came after
in the 1980s, still required the speaker to speak
very slowly and clearly. But then in 1997, a
speech recognition programme called Dragon
Naturally Speaking was released. After training
the programme for forty-five minutes, it would
recognise your natural speech (around one
hundred words per minute).
The technology has continued to develop to
the voice recognition features we have on our
mobile devices today, allowing us to call friends
by saying their name, easily search when we’re
on the go, dictate text messages and more.

160