|
Software
PC /
Windows programs
Macintosh
programs:
Hardware
Desktop
versus portable notebook
System
requirements
Microphones
- - - - - - - - - - - - - -
- - - - - - - - -
Updated July 2005
Software
Speech recognition programs have
come and gone over the years. The two programs that have been
around the longest are Dragon NaturallySpeaking and IBM ViaVoice.
Both programs are currently sold by ScanSoft (July 2005).
In general, speech recognition involves the ability to match a voice pattern against a provided or acquired vocabulary.
Currently, full dictation speech
recognition programs are 'user dependent' meaning that
each user is required to train their voice on the system, or
enroll through training so that the computer recognizes their
own specific, unique speech patterns, pronunciations and
intonations. Programs normally provide a large built-in
vocabulary which can be modified and developed further depending
on each individual users speech patterns, frequently used words
and language.
Programs that are currently available and viable programs for
the educational field for students with disabilities (July
2005) are:
| PC /
Windows programs
|
Macintosh
programs:
|
top
- - - - - - - - - - - - - -
- - - - - - - - -
Dragon NaturallySpeaking 8 from ScanSoft
Reviews of speech-recognition programs by the computer industry
have consistently rated Dragon NaturallySpeaking higher than any
other speech-recognition program. It has always been rated
highly in terms of both accuracy as well as user interface.
In the educational field, NaturallySpeaking has been the program
of choice for people working on a PC platform. Initial accuracy
percentage rates of 97 - 99% are possible following initial
training and practice. A rate of over
100 words per minute is achievable with work and
perseverance. Initial voice enrollment (training
your voice file for the first time) takes less than 5 minutes on
a fast machine.
Versions:
NaturallySpeaking comes in three different versions:
standard, preferred and professional.
- The
standard version
provides full capability for dictation and creation of
text.
- The
preferred version
is, in most cases, the most useful version for students
with disabilities. Apart from being able to
dictate text and have text appear on the screen, it also
includes "Text-to-speech" screen reading capability as
well as dictation playback. This means that the students
can hear his own voice repeated as it was recorded, as
well as hear the text which was placed on the screen
during dictation read with a synthesized voice. These
features are particularly useful for students with
learning disabilities, difficulty concentrating, those
who have difficulty editing their work, and students
with visual problems. This version also has some
limited macro recording capabilities, which allows for
larger amounts of text to be placed on the screen simply
through the dictation of one or two keywords.
- The
professional version is
a full-featured, extremely powerful program which allows
for extensive customized macros, completely hands-free
use and also recording of the entire audio recording of
the whole document for later editing. In most cases, the
professional version is too expensive and sophisticated
for students. However, some students with severe
physical disabilities who require hands-free access as
well as higher cognitive functioning students who can
make use of extensive macros for improved efficiency may
benefit from the professional version.
Click
here to view a detailed feature comparison matrix
available on the ScanSoft web site.
Other important features
of Naturally Speaking of importance for people with
disabilities:
- Navigate the computer
desktop by voice
- Filters out inadvertent
fillers and sounds between words.
- Mouse control and working on
the web
- Macros allow for blocks of
text and pictures to be inserted with a quick voice command
- Import / Export user profile
and voice files - for use over a network
- - - - - - - - - - - - - -
- - - - - - - - -
top
See the following reviews of
Dragon NaturallySpeaking 8
QPointer
Voice
QPointer is primarily a voice operated navigation tool.
It does allow for dictation, however the accuracy and
efficiency of this dictation is not anywhere as good as
Dragon NaturallySpeaking. QPointer allows the user to
use any combination of mouse, keyboard and voice
controls. Users can use voice input instead of mouse and
keyboard to navigate the computer as well as enter text.
This program was developed specifically for people with
disabilities.
IListen from MacSpeech
This is one of the few options available for Macintosh
uses. Compared to the other option for Mac, IBM ViaVoice,
iListen is more accurate. It is not as sophisticated and
does not work as well as Dragon NaturallySpeaking, which
is only available on PC. iListen became available at the
end of December 2000 and is a relatively new program. It
allows dictation, transcription, formatting and speech
navigation into any application. It will transcribe text
from audio files and supports the Olympus series of
digital voice recorders. Version 1.6.8 (released in May
2005) has special features for Tiger. iListen
cannot be used in hands-free mode.
MacSpeech is the only company currently developing
dictation software for the Macintosh.
IBM ViaVoice for Mac OS X
from ScanSoft
Dictation accuracy in ViaVoice for Mac OS X. is lower
than the accuracy achieved with iListen. In addition,
users can dictate and edit only within the small word
processor provided with IBM ViaVoice. Overall, this
program is more limited than iListen.
- - - - - - - - - - - - - -
- - - - - - - - -
top
Both Macintosh Mac OS X. and
later and Windows XP computers come with some form of
built-in speech-recognition. Command and control
capabilities are part of the Macintosh operating system.
Windows XP speech recognition allows for dictation.
Neither of the speech-recognition capabilities of these
operating systems are truly functional for full
speech-recognition required by students with
disabilities.
- - - - - - - - - - - - - -
- - - - - - - - - A
General Survey of SPEECH RECOGNITION PROGRAMS
(December 2004) - A good overview of the
current status of speech recognition programs.
General Voice Recognition software
reviews (October 2004)
Reviews top
- - - - - - - - - - - - - -
- - - - - - - - -
Hardware
(as of July 2005)
Desktop computer
versus portable notebook computer
System
requirements for NaturallySpeaking v8
Desktop
computer versus portable notebook computer
When considering the use of speech recognition in a school
environment, the logistics in terms of hardware, location of
hardware, student mobility throughout the school building and
student use of the system for both school work and homework,
need to be taken into consideration.
A couple of options can be considered in terms of hardware
requirements: In some cases, the most effective system is a
laptop computer which would allow the student to use speech
recognition both at home and at school, while still working on
the same voice files. Effective speech recognition is based on
the user's ability to continually build on voice files. Each
time the program is used, additional words are trained, and the
vocabulary and recognition of the user's voice improves with
each use. This means that it is essential for the user to work
on the same voice file at all times, both at home and in school.
This can be effectively achieved with the use of an individual
laptop which the user uses both at home and school. The use of a
laptop or notebook computer is especially worthwhile if the
student is using other specialized assistive technology
applications to, in addition to speech recognition. This will
allow the students individualized computer set up to be with the
student in all work environments. In some school environments,
where assistive technology applications, hardware and software
is not readily available in many different rooms in the building
, it may be more cost effective to provide certain individual
students with laptops.
In other instances however, some students are able to make use
of a USB or memory key to transfer voice files from one machine
to the next. In this case, the student would need access to a
number of different machines which have the same version of
NaturallySpeaking loaded. If a school building has
NaturallySpeaking loaded on multiple machines in various places
in the building, in places where the student would need to work
with NaturallySpeaking, the transferor of voice files with a
memory key may work effectively. This would only work for a
student who exhibits good work skills and responsibility, as
well as the organizational skills required to keep track of
where the most current voice files were located.
There are benefits and negatives associated with both the use of
desktop versus laptop or portable computer systems for speech
recognition and all assistive technology. The provision of
specific hardware is a highly individualized process based on
not only student needs, but also the resources and equipment
available within the educational environment.
See the 'Hardware
compatibility list' for Dragon NaturallySpeaking which will
give an indication of the types of hardware which works best
with the program.
top
System requirements for
NaturallySpeaking v8
These are the
requirements posted on the ScanSoft web site. They should
be considered absolute minimum requirements, on a machine
where nothing else but NaturallySpeaking will be running.
Higher system specifications are preferable for efficient
running of NaturallySpeaking together with other programs.
Newer, entry level systems being sold (as of July 2005), have
processor speeds of well over 1GHz. As such, it is clear
that system requirements and needs for speech recognition is no
longer the cost issue it once was with processor and memory
hungry speech recognition programs. Currently, the
hardware has developed to the stage where even entry level,
cheap computers will run the software.
- Intel® Pentium® III / 500
MHz processor (or equivalent AMD® processor)
- 256 MB RAM
- 500 MB free hard disk space
- Microsoft® Windows® XP (SP1
or higher; includes XP Tablet Edition) Home and
Professional, Millennium, 2000 (SP4 or higher)
- Creative® Sound Blaster® 16
or equivalent sound card supporting 16-bit recording
- Microsoft® Internet Explorer
5 or higher (free download available at
www.microsoft.com)
- CD-ROM drive (required for
installation)
- ScanSoft-approved
noise-canceling headset microphone (not included with
upgrades)
- Speakers (required for
playback of recorded speech and text-to-speech features)
- A Web connection is required
for activation (although this can be done manually on
another computer connected to the web)
Microphones
Remember – better computer and
higher quality microphones will always result in better overall
performance and recognition.
Some excellent
articles about microphones specifically for speech
recognition.
top |