Dragon Dictation: Mythically Good Speech Recognition

Dragon Dictation is an app for the iPad that allows you to speak into the iPad, and have the words automatically transcribed for you into text. It wouldn’t be correct to type this article then, would it?

So what you are reading has been completely and automatically transcribed by this application whilst I sat in cars, cafes, and coffee shops, chatting away to my iPad.

Embarrassing – yes. Enlightening – absolutely! Let me share with you what I found…


On opening Dragon Dictation.

Dragon Dictation utilises the microphone and processing power of your iPad to turn spoken words into text. It also requires a connection to a remote server for assistance.

Whilst it needs an internet connection, operation over 3G is perfectly fine as only small amounts of data are uploaded and downloaded. This whole article uploaded less than 3MB, and downloaded just under that.

One operational point – I had to remove my smart cover in order to access the microphone clearly. You may also have to remove your tiger-pattern plush velvet cover too, or whatever you bought, in order to be heard clearly. Another minor point is that the “Speak here” advice that points you to the location of the microphone, still points to the original iPad location, not to the centralised iPad 2′s microphone.

The interface is minimal, with only a central record button, + symbol to start a new session, bin to delete notes, and some export and settings options under the final “arrow pointing to line” symbol (what does this symbolise? Answers in the comments below….):

Minimalistic menu..

Minimalistic menu.

The note button will take you back to previously recorded items, or you can rotate to landscape to see brief headers in a left-hand pane. It would be useful to have the record button as one of the hardware buttons. It would also be useful to have Dropbox integration, although there are some useful export options already included.

For simple uses such as dictating a short piece, Dragon Dictation is excellent. It can also be used to dictate longer items, such as an article or several paragraphs, and I found it helpful when first starting out to initially record for only about 10 seconds or so before pausing to check the text. This saves wasting time by speaking for a very long while, only to realise that the app is having difficulty with your voice, perhaps because your glove is over the mic, or (and) you’re munching on a cheese toastie…

To be able just to speak to the iPad at a reasonable speed and have the words appear in text is quite simply magical to do and see. Dragon Dictation does this excellently, and I feel that it’s a great use of the iPad’s processing power and speed. You can then decide what to do with your finished transcribed text…

A Multitude of Uses

What can you do with your transcribed text from Dragon Dictation?

  • Upload a tweet direct to Twitter.
  • Facebook updates are a breeze.
  • E-mail your spoken masterpiece.
  • Copy and paste into other apps, such as PlainText.

For quick ‘off the cuff’ remarks, Dragon Dictation can rapidly get them up onto your favourite social networking site…

Coffee :)

Speaking a Facebook update, smiley and all, is a breeze...

One of the things that is a bit frustrating is touching the red record button to start and stop recording. It takes a little bit of practice to hit it just right every time, and would be more suited to a re-programmed hardware button.

Should you need to make changes to your spoken text the iPad on-screen keyboard is always handy and can pop up on a touch of the keyboard icon which is always available on screen. The built-in text editor is basic but functional, it gets the job done. Unfortunately, you can’t paste text into the app, to do some vocal editing, it’s just not the focus of the app.

Once you’ve finished recording your speech the iPad goes into processing mode converting your spoken audio into text. It is quick, and suprisingly accurate for a first time use. The app learns over time to recognise your voice and the way you say things, so accuracy should improve the more you use it.

Self-Consciousness

Of course one of the main objections to using something like Dragon Dictation is the self-consciousness of speaking out loud to a machine, even in a car or in another private space if you don’t fancy braving your local coffee shop for a cosy chat with your tech. For me this distraction is a small price to pay for rapid text input in an accurate manner, but it probably pays to be a little extrovert or not minding attention…

With the introduction of Siri this kind of talking to your devices will hopefully become more socially acceptable and normal.

One of the trickier things to get used to is simply speaking punctuation. Thankfully there is a quick-reference guide under the ‘i’ at the bottom right of the app:

Punctuation tips

Punctuation tips available under the 'i'.

Of course accurately transforming your spoken word into text is just half the battle. First you have to have something good and useful to say. This takes a little practice, formulating sentences in your mind before speaking them. Arguably, it makes for a more disciplined, perhaps clearer, writing style.

As you’d expect Dragon Dictation has an inbuilt dictionary to help with individual word correction, and copes with many unusual and unexpected words and phrases.

For best results, though, find a quiet place to sit and speak clearly – the less background noise there is, the better the results.

It also helps if you have clear diction and say-each-word-very-distinctly. This saves a lot of typing, or checking afterwards. What is perhaps even more suprising, though, is that you can just chat away at a normal conversational speed, and the accuracy stays respectably high. Here is an example of the famous Shakespeare speech from Hamlet, To be, or not to be…, spoken at my normal conversational speed:

To be accurate, or not to be...

To be accurate, or not to be...

And, to conclude on a jolly note, the application also has a very cool party game. Simply have someone say a well know phrase or song into Dragon Dictation, at a decent speed, and watch with amazement at how much it gets, or laugh at its choice word replacements. Try it with your friends! You know you want to…


Summary

Converts spoken speech to text automatically, and with impressive accuracy. And it's free!

8
  • Don Seher

    When I need to create computer instructions for the Mac users on campus, I set up my iPad, start Dragon Dictation, and verbalize what I’m doing as I go through each step on my iMac or MBP. In a matter of seconds after I’m done, the instructions are translated to type and ready for me to place in Docs To Go to fix up. After fixing up, I use FileBrowser to place the instructions on a server for access by the Mac users.

  • Cal

    Can you use headsets with the built in microphone to dictate, or do you have to use the iPad’s mic?

    • Joel Bankhead

      Hi Cal, yes you can use the built-in headset microphone – no problem!

  • http://twitter.com/DarylGriffiths Daryl Griffiths

    Mark, you mention having to hit the red button in the middle of the screen, but this isn’t actually necessary – you can tap anywhere on the screen for it to both stop and start recording.
    You don’t seem to have mentioned, and may not therefore be aware, that sometimes if the app isn’t entirely sure which word you intended to use – and thinks there may be possible alternatives – it will list those if you tap the word in question once it has been converted to text.
    I haven’t found any great difficulty in getting the app to understand me, but I am used to using the DragonDictate software on the desktop, so probably have something of an advantage, although I do occasionally find myself slipping into trying to give voice commands to the app which it doesn’t understand!
    It’s not necessary to edit text by using the keyboard as you can click and select the word or phrase and then click the red record button again and overwrite the earlier dictation.

  • Everbloom

    In my circle, we use Dragon Dictate as a kind of running joke. Maybe it can cope fine with Americans, but it is absolutely hopeless with Australian English. It’s easier, faster and more accurate to type. Even on the iPad and iPhone.

    Sometimes I redownload it thinking it can’t be as bad as I remember, but it always is.

  • Cal

    Thanks :)

  • Marc

    Great article (and site)!
    Dragon dictation is asking me for my country at setup.
    Does it mean that I won’t be able to use it in english later if I set it in french the first time I use it?
    I couldn’t find informations online.
    Thank you!

  • NK

    i searched this app even before the siri launch. But to my dismay, it isn’t available in my country’s appstore(India)

  • Joe Murphy

    Thinking of getting an Ipad for my elderly mother to use, along with Dragon Dictation, in church. She cannot hear and this thing appears to be the perfect solution.

    My question: is there a time limit on the amount of audio that it will handle, other than the amount of space on the Ipad storage?

  • David Phillips

    I had a friend ask me if he, a dermatologist, could upload a dictionary of dermatological terms to the iPad so that they could be understood when dictating patient diagnoses or examinations. Is this possible?

  • PM

    DragonDictate enjoys greater popularity than it deserves. Because there are so few alternatives I have persisted with it for years because of a shoulder problem and difficulty with typing. The problem is the software is incredibly resource hungry and prone to compatibility problems with many other pieces of software in a multi-tasking environment. The people marketing dragon focus more on marketing then product development and improvement. They have been saying for years you can achieve 99% accuracy which is a theoretical possibility only achievable if you read a set piece of text from a book. In real-world conditions where you are trying to dictate into Word or Outlook it can be very problematic with freezing and compatibility issues with other software running. Also the AI or artificial intelligence is very frustrating because even when you get good recognition it often inserts an incorrect version of your text into the sentence or paragraph you are working on. Once you need it to dictate into large documents in Word et cetera you are really pushing the boundaries. It is extremely difficult to get it to work in Microsoft WebOutlook. It simply has difficulty in any HTML environment. I know I sound negative but I really hope in time there is more competition in this area because voice dictation is the future. The support from Nuance who distribute the software in Australia and possibly elsewhere is really not very good. Unless you are registered they won’t speak to you, fair enough, but the support of the product expires after a short time anyhow. There is a paucity of information available online or in discussion forums on how to fix common problems. The general response of the company is the software is incapable of doing this or that. So there seems to be a lack of commitment to continual improvement and development. This is despite all the marketing hype to the contrary. There really has not been significant improvement in the software between versions for some time. It’s interesting to see that the same problems are being imported into new applications such as with iPad. I assume that much of their profitable selling is to specialists/professionals in legal and dental where they sell expensive databases to support the software. Many of the average punters I have spoken to end up not using the software very much. For people like myself who use it extensively it has become very frustrating and we eagerly await an effective competitor to come into the market.

    • Katie D

      Thank you for the detailed information! I am a high school special education teacher stumped at how to work with a student at transferring the knowledge in his mind to his paper. I thought this would be perfect for him, however it is extremely inaccurate and unusable in the education setting. If you have any suggestions please post as all suggestions are greatly welcomed!
      Cheers!

  • Richard

    I am late deafened and sometimes use my voice but can’t hear other people. Can I stick an iPad in your face and a dialogue take place or does DD require voice profiles to be created like other voice recognition softwares?

  • Andrew taylor

    Can i useDragon to make notes while reading an ebook on the iPad?

theatre-aglow
theatre-aglow
theatre-aglow
theatre-aglow