Page 156 - Handbook of Electronic Assistive Technology

144 HANDBOOK OF ELECTRONIC ASSISTIVE TECHNOLOGY

Amazon Echo, Google Home and Apple HomePod are a variation on the personal digi-
tal assistant theme. They offer similar functionality to the smartphone/tablet tools, but in
a dedicated hardware package. In addition, they can control home automation systems
and therefore offer an option for environmental control; SmartBox offers a grid set so that
users with communication impairment can use their voice output device to control the
Amazon Echo (Smartbox AT, 2017).

Dictation

Suppliers of speech recognition technologies for dictation claim rates of up to 120 WPM,
although actual rates are usually considerably slower when taking into account time to
compose, review, correct and edit the dictated text. In a recent systematic comparison
of studies of assistive technology access interfaces, speech recognition is reported as the
quickest, averaging 15.4 WPM (Koester and Arthanat, 2017).
Most modern operating systems provide a speech recognition dictation functionality:
Siri Dictation on iOS and MacOS; Google Voice Typing; and Microsoft Dictate, for example.
Again, an internet connection is required for these systems. These built-in dictation tools
do not offer text editing or correction, and so the user must have some other method of
selecting and editing misrecognised text.
Dragon NaturallySpeaking and Dragon Dictate for Mac, in contrast, do provide control
with speech to correct misrecognition errors and to edit and format text by voice.

Computer Control

The digital assistant speech recognition systems have limited functionality and do not
provide complete control over the device. While they may be extremely effective at what
they do for an individual user of electronic assistive technology, in most cases another
method of access will also be required.
A greater level of device control is provided by Windows speech recognition and partic-
ularly by Dragon NaturallySpeaking, which provides dictation, text formatting and editing,
integration with Microsoft Office, access to menus and dialogue boxes and mouse control.
Windows and Dragon NaturallySpeaking do not require internet access to function, which
can be an advantage for some users and applications.

Alternative and Augmentative Communication
Speech recognition has been proposed as an access method for personal communication
for people with dysarthric speech. VIVOCA is intended to accept disordered speech from
the user and ‘translate’ it into synthetic speech output (Hawley et al., 2013).
Microphones

Sound quality and consistency are important for accurate speech recognition.
Microphones built into smartphones and tablets are usually of high quality, but accuracy