Speech Recognition

Speech Recognition Software (SRS)

edited by Malcolm Litten.

SRS has been around for over 20 years now.  It has had a mixed reputation, partly due to its availability in the early stages of development and partly due to successful use being dependent on the user’s willingness to learn how it is best used and persist through any initial frustrations.

Speech pictureThere are several versions of SRS available.  All of them require careful voice training by the user.  

There is a free version built-in to Windows Vista, 7 and 8, which can be found in the Ease of Access Centre. A commercial programme, SpeakQ, has been developed using this version, and is designed to assist poor readers through the training process. Read & Write Gold has a link to the free Microsoft SRS. 

The main commercial program is Dragon NaturallySpeaking for use on PCs. For dyslexic users, Dragon NaturallySpeaking is currently much the most user friendly. 

There is an entirely different program for use on Apple computers, despite the similarity in name, called Dragon Dictate

There are free app versions of Dragon, called Dragon Dictation for use on tablets and smart phones.  These require no training.  Conversion to text is done remotely online and there are time limits to any one piece of dictation.

iPads (3 or later) now have built-in speech recognition. This is activated by switching Siri on. Then a key on the on-screen keyboard has a microphone on it. When this is tapped, you can dictate text. Like the free Dragon app, this requires no training of the software to recognise your voice. However, this means that if recognition is consistently poor there is nothing the user can do to improve performance.

SRS is now being very successfully used by many people with dyslexia.  However, like any form of help, it will not suit everyone, so below are a list of pros and cons for you to consider.

Back to the top

Good points:

  • No typing skills or spelling knowledge are required to create text.
  • So all the user’s attention can be focused on thinking about the content.
  • It can save time for slow typists.
  • Dragon NaturallySpeaking (Premium or Professional only) includes recording of the dictation. This enables the user to hear what they actually said where an error in recognition occurs.  It also has built-in text-to-speech in all versions enabling the user to hear the text that has been produced.
  • These features can greatly improve the user’s proof-reading success.
  • It is possible to dictate in any space where you can type, so is equally useful for emails, texts and social network sites as well as word-processed documents and spreadsheets.
  • There are hundreds of commands that will allow the user to format, control the mouse and even utilise macros to complete lengthy routine tasks with a single command.  But almost none are essential for successful dictation.
  • Depending on the user, SRS may lead to far more being written, a greater range of adventurous vocabulary being employed, text being readable for others for the first time ever and learning being enhanced by the active recording of knowledge and understanding.
  • Dictation on a digital handheld recorder can be transferred to the computer and transcribed – Dragon NaturallySpeaking -Premium or Professional only.
  • If a candidate is entitled to a scribe in examinations, SRS can be used if it is the candidate’s normal way of working.
  • Concerns about gender, age or accent are unfounded.

Back to the top

However…

  • Do not expect perfection.  The claim of 95 to 99% accuracy made for Dragon NaturallySpeaking is not hype, but 95% = one word in twenty being wrong.
  • Fluency in delivery helps the accuracy of recognition, so halting or stumbling delivery increases errors.
  • Clarity of delivery is sometimes an issue due to speaking too rapidly or ‘swallowing’ small words.
  • The initial training of the software to recognise the user is now quite quick (maybe 30 minutes for a slow reader) but the real time commitment is in becoming an experienced user, willing to correct errors of recognition in the recommended way.
  • The headset is the vital link between user and computer.  In some laptops, better results are obtained with a USB input on the headset.  If necessary, adaptors can be bought separately that cost about £12.
  • Generally speaking, success is more likely with adolescent and older users.  However, examples of success with younger users are not that rare.  Degree of need and personal qualities will both play a part in success or abandonment.
  • Some users will benefit from preparation of ideas and planned structure before dictation.

You should always check system requirements for any software.  It is unreasonable to expect successful performance if you try to cut corners on specifications.

SRS has proved life-changing for some users with dyslexia.  It is not a cure-all.  If it matters to you, find support from an experienced user or pay for training.  Do not expect to just install it, don a headset and chat away!

See dyslexic experiences of Using Speech, S R parental advice and S R questions.

Back to the top

© B.D.A. New Technologies Committee. July 2015.
Copies of this page may be made providing it is unchanged and the source is acknowledged.

2 Responses to Speech Recognition

  1. Pam Lore says:

    That is really helpful. Does anyone have experience of the Windows 7 free app?

    • Mick says:

      Yes. Its available in the “Ease of Access Center”. Like Dragon it needs practice and I would advise downloading the commands.. If you want the typed text read back, select “Narrator”. It’s also available Windows Vista, Windows 8 and Windows 8.1, find “Ease of Access Center” and make your selection. You most certainly don’t need to fork out 12 quid for a headset, since if you search there’s lots of cheaper ones on the web including a throat mike headset for around half that price. The NTC seem to be fixated by Dragons. There’s no mention of Apple’s Siri running on iOS7 devices either. Whichever way you go Dragon or MS Windows there’s not much difference in performance in my experience as long as you don’t try to dictate too much in one go. To get them to perform accurately requires a lot of work. Remember also that these programs learn your individual style and can come out with something completely different in text if you have a cold or sore throat

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s