Using speech

Rob Wheeldon’s experience with Speech Recognition. May 2012.

I am writing this with my chief study aid called Dragon NaturallySpeaking. I’m currently training it, and with a strong north-west English accent, this is not always an easy task! However it is considerably easier than touch typing which I manage to do with one finger. I never learned to type at school; most boys didn’t; we did wood and metal work instead in the 80s. This seems an incredible oversight in this multimedia day and age. Although I’d like to be able to type, realistically I never had the time to do this at University whilst still covering my study load. Dragon NaturallySpeaking saved me a great deal of time and effort and I never would have completed my dissertation without it. Dragon NaturallySpeaking was not supplied to me by the University but was rather my own purchase (a purchase well worth making). Nobody saw the information age coming and the usefulness of having typing skills.

Dragon NaturallySpeaking is advertised on television as a
dictation device. You wear headphones and talk into a microphone. You try to talk as naturally as possible and most of the time it gets the words right (I’d say about 75% of the time in my case but with training and continued usage this does improve). However it does get a little frustrating when it continually gets a word wrong at which point I switch off the microphone and type the word in (this only happens with certain words). This most recent version of Dragon NaturallySpeaking cost me £50 from Amazon which I consider to be money very well spent.

When spending a long time dictating I would recommend the occasional short break as it can be quite mentally tiring if you are unused to it. Dragon has both improved in accuracy and dropped in price since the first version I used six years ago. If you are dyslexic or have a dyslexic child who is struggling at school this is one of the main products I would recommend. Having said that dyslexia is a syndrome and not a medical condition and it may not prove useful for all dyslexic individuals.

Dragon. May 2011.

There have been recent updates of Dragon NaturallySpeaking and Dragon Dictate for Mac. They have become a little more dyslexia-friendly by presenting the training passages for reading in much shorter chunks.  You can also print out these passages for pre-training practice.  The different price levels have been renamed ‘Home,’ ‘Premium’ and ‘Professional.’  ‘Premium’ is best for dyslexic users as it includes the facility to record your dictation and hear it back as required in addition to the text-to-speech engine that reads the text on the page.  Don’t forget you can get reduced price copies of either program if you are in fulltime education.

See current version of Dragon NaturallySpeaking by Nuance.

B.D.A. Contact magazine. September 2010.

Speech Recognition Software: Dispelling the Myths.

Rhodri Buttrick.

In April 2007, Rhodri Buttrick made a video, on behalf of the BDA, about his use of technology which was shown at an e-Forum conference in Westminster. It demonstrated his use of some specialist software, in particular the use of the speech recognition program Dragon NaturallySpeaking. In 2008 he gave an address, on the same theme, to an EC ministerial conference on e-inclusion in Vienna.  Rhodri began using voice recognition software, with his parents’ support, when he was nine years old. Today he is at university and continues to use the technology to help him with his writing just as he did throughout his time at school. In this article, he describes some of the common misconceptions about its use based upon his own experiences.

I want to clear up some myths and misconceptions surrounding speech recognition and shed light on a few things you might not know about it. It seems whenever I mention that speech recognition is my main method of work, there is always some clever clogs who says, “It’s rubbish” or, “I tried that and it didn’t work properly”. The reason most people have this opinion is because many who have tried out the software did so back in the 90s. Unfortunately, this concept of the 90s “Ghost of Speech Recognition Past” is still hanging around, much like 90s boy bands!

Due to my dyslexia and dyspraxia, I find handwriting impossible. This made essay writing at school incredibly frustrating as I wasn’t able to get all my ideas down and subsequently scored terrible marks. My parents did some investigating and found out about speech recognition software. Thank goodness they did, as without it I would not have passed my GCSEs let alone be studying philosophy at University. It made me independent and I do not have to rely on an amanuensis. It was the solution I was looking for, but it was certainly not easy and I will talk about the difficulties I faced, as a dyslexic, learning to use speech recognition software.  I had tried learning touch typing but this was a long and gruelling process and I could only type short paragraphs before the effort became too painful.

I began by using Dragon NaturallySpeaking Version 5. When you first install speech recognition software you need to complete the training process for the computer to learn your voice. This involved reading extracts from books into the computer. I was nine years old at the time and, being severely dyslexic, I could barely read. My mother had to sit behind me, whispering the text into my ear in short phrases, which I repeated into the microphone I was using. It was hard work, but I persevered as this was the last chance for me. With my parent’s support, I adopted the “good practices” recommended by the software supplier, such as correcting any recognition errors as I went along using the “Correct That” function, rather than overtyping. The result was I achieved a recognition level of over 98%, talking in a natural way, at normal talking speed. Many people give up on speech recognition because they don’t get into good habits. If you find a word is consistently appearing wrong, you should use the “correct” function to type in what the word is meant to be. By doing it this way, the software continuously learns thus making the recognition even better.

Speech recognition is actually quite a bizarre skill and is very different from dictating to a human being. Speech recognition can only transcribe what it “hears”, so if you slur your words or leave some out altogether, you are going to have accuracy problems. As long as your pronunciation and diction is consistent, the software will grow used to the way you speak.

The following are Rhodri’s myth breakers and recommendations:

  • You can use Speech Recognition Software with almost any application.
  • If you have difficulty manipulating a mouse because of a physical disability, you might find browsing the Internet difficult. Dragon labels each of the hyperlinks with a number so all you have to do is call out the number and it clicks on the link for you.
  • If you receive Speech Recognition Software as part of the Disabled Students Allowance, you can also ask to receive training in its use.
  • The accuracy of speech recognition improves if used properly.
  • It is possible to create specialist vocabularies to suit your particular work.
  • If there is a particular piece of text you find yourself having to write frequently, such as your address when you write a letter, Dragon NaturallySpeaking allows you to have a code word for it. For instance, there is the option to program it to type out your entire address every time you say the phrase “my address”.
  • It is a common belief that Speech Recognition does not work for regional accents. This is not true. What matters is pronouncing words consistently and not leaving out sounds.
  • It is also a myth that children can’t use Speech Recognition. I started using it when I was nine and I didn’t even have to do much retraining when my voice broke! However, anyone teaching children MUST be competent in using the software themselves.
  • When using it you don’t have to talk REALLY LOUD! When I was at school this myth was one of the reasons teachers were reluctant to let me use it in the classroom. However, I am using the software right now and I’m talking barely above a whisper. When using it in lessons, I was certainly never any louder than the guys gossiping at the back of the classroom!
  • It is a myth, that when you are dictating, the room has to be totally silent. This was a concern I had when using speech recognition in a busy classroom. However the microphones used for speech recognition generally tend to be proximity microphones which only pick up noise close to. Unless somebody’s being particularly loud next to you, the odds are, recognition will not be affected.
  • Training is now very quick. With the latest versions of software Dragon NaturallySpeaking only takes about 30 minutes to do initial training.
  • You do not have to speak each word individually like a robot. In fact it works much better if you speak as naturally and as fluently as possible.
  • You do not have to speak really slowly. In fact, it’s better if you speak at a normal pace. If you pronounce each individual word properly, you can speak as fast as you like.
  • It is a myth that you are not allowed to use speech recognition in an exam. Access regulations for GCSE and A levels do allow for its use. Universities also allow it. Schools and universities sometimes make a meal out of this one but persevere, as it is permitted but not many people know this.

One final note: Speech recognition is recommended as a key “access” arrangement in schools and universities.  However, it is important to remember that it is not a “fix all” solution. It may have been the right approach for me but you must consider very carefully, after trying it, whether it is really the right approach for you.

Rhodri’s original presentations on speech recognition are:
Demonstration of voice recognition for Westminster E-forum and
e-Inclusion speech at European Commission’s ministerial conference.

More information can be found on
Rhodri’s blog and his website.

Editor’s Note: Rhodri uses Dragon NaturallySpeaking Preferred with the addition of a noise reducing headset. Version 13 is now available.

See main page on Speech Recognition, S R Parental advice and S R questions.

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

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