Does Speech Recognition Software
Really Work?

image of a hands-free headset

One of my favorite posts from around the web last week came from our own Associate Editor Jon Morrow. He recorded a 20-minute video post for Problogger about how he works with speech recognition software to do all of his blogging.

I do an awful lot of writing every week, and I’ve been thinking about trying speech recognition out in order to speed up the process. But like most people, I was afraid it was going to be more trouble than it was worth to get it working.

Jon’s video made me realize how simple (and inexpensive) it will be for me to make it happen.

Because it was a pretty content-rich video, a lot of folks took a quick look and bookmarked it, thinking to come back to it when they had a little more time. So what better way to spend the Friday-after-a-holiday than eating leftover turkey sandwiches and watching a great how-to post?

(If you’re not in the States, you can re-create the effect by overeating wildly today or tonight, drinking just a little too much, pounding down four desserts, having three arguments with your extended family, and then watching the video tomorrow.)

The highlights of the video for me were:

  • The quick-to-install (and cheap) piece of hardware that lets the software actually understand what you’re saying.
  • Jon on video! Jon and I have spent a lot of time on the phone, so I’ve gotten to know him fairly well. Getting to hang out with him for a few minutes via video was great, he’s a fascinating guy with a lot to say. (The guy can say more with his eyebrows than most people can with a 100-item list post.)
  • The one-stop resource to find the right mic and hardware for your setup.
  • The live demo showing exactly how Jon uses the software to manage his business and blogs.
  • The comical notion that penny-pinching Jon will ever buy a Mac.

I recommend you check it out, I found it tremendously useful:

Speech Recognition for Bloggers: The Ultimate Guide

About the Author: Sonia Simone is Senior Editor of Copyblogger and the founder of Remarkable Communication.

Print Friendly

Smarter is Better Solutions for Smarter Content Marketing

Here’s what we’ve got for you:

  • 15 high-impact ebooks on content marketing, SEO, email marketing, landing pages, keyword research, and more.
  • A 20-part Internet marketing course that lays out a comprehensive path for your own online strategy.
  • An organized reference guide to the “best of the best” of Copyblogger.com, and how it all profitably fits together.
Free Registration

Take The Conversation Further ...

We'd love to know your thoughts on this article.
Meet us over on Google+ or Twitter to join the conversation right now!

Comments

  1. My experience with voice recognition:

    I’ve been using Dragon Naturally Speaking. Getting a different headset made a big difference. With my Logitech behind-the-head set (about $15), the program is doing a better job of recognizing my speech, as well as distinguishing commands. I still get frustrated with it for not recognizing what I think are simple words, even though I make a point of using excellent diction.

    I also got better results when I switched to the slowest speed for conversion; it takes longer for the words to show up on the screen, but it allows the program to run through more vocabulary options to get the best match. I love the fact that I don’t have to say “period” at the end of a sentence and “cap” at the beginning of the next sentence.

    Dragon isn’t living up to its promise of usability in Word, though. That’s a big disappointment. I have to paste text into Word.

    I wish you success with your adventures in speech recognition.

    Note: My website isn’t up yet. I came here for guidance from you, and I am finding what I need. Thanks!

  2. Thanks for reminding me about this. I used it for a while a couple of years ago, but was put off by the length of time it took for the programme (Yes, I’m English!) to learn the sound of my individual voice and my various pronunciation ‘peculiarities’. Hopefully the software in its latest versions will have improved its learning speed. Yes, I think I’ll give another whirl.

    By the way, while I’m here, thanks for an excellent blog.

  3. Great post and the first comment really helped me out alot. I have been trying to get this voice recognition to work because I have a book I want to write but hate typing. thanks for this great post. Larry

  4. hi guys, I got Dragon this week after persistent arm pain that has all my doctors baffled. The only thing that relieves my pain is tons of yoga, vacations away from the PC, and a short not typing for long periods of time. I have spent a total of about 6 to 8 hours of both training Dragon and using it, and I think it is really worth the effort.
    I noticed that someone mentioned problems with Microsoft word. I have not noticed any problems using it with word. I do however have a hard time using it with Gmail in the browser.
    I was also inspired by that video when I first watched it. Isn’t it amazing that there are people who are unfortunately paralyzed and have no other option but to use speech recognition software, and that technology enables them to do so and still achieve great accomplishments using their computers?
    By the way, this entire comment was typed without touching the mouse or the keyboard!

  5. Of course in re-reading, I see that Dragon made assumptions. I made I meant “a while” not typing…..!!! Forgive us :)!

  6. I thought two points in Jon’s post were particularly interesting. One was that the newest version of Dragon worked so much better than previous versions, and the other was that a cheap USB sound pod would make such a big difference in getting the software to recognize words. That, and using KnowBrainer.com to get the best combination of mic/sound pod. (Although unlike that cheapskate Jon, I’ll buy my stuff there rather than looking for a better price elsewhere.) ;)

  7. Great post! I’ve often wondered if speech recognition systems were worth the trouble; now I know the answer.

  8. Back in the day (about 5-6 years ago) I had the hope that all my writing could be done with speech recognition software. But then I couldn’t even get past the part where I was training the software to understand my voice.

    I considered trying again, but I don’t think I even speak as fast as I type.

  9. Hi Sonia

    I have been using Dragon for a couple of years now and, yes, it takes a while to tame and train the dragon. But overall, when my shoulder starts hurting and feeling the strain of being on the computer for long periods of time, it is a relief to put the headset on and talk instead.

    It can be quite entertaining as well, to look up and see what the dragon THINKS I’ve been saying.

    Luv the dragon
    Kathleen

  10. Speech recognition software surely do not work for non native english speaker. Yeah– diffrent prounouce will give diffrent output, don’t it?

  11. Dana, since the software learns your pronunciation, I think it would probably work fine. As long as you pronounce words consistently, I think it would adapt. Might take some time to get the software in sync with your voice.

  12. In my day job, I’m the archivist for a large spoken word audio collection. A common question I get is: “Can’t you digitize everything and generate automatic transcription?”

    Well, not exactly. Software like dragon can learn to recognize one person’s voice, but not multiple speakers with muddy sound and possibly various accents. It’s difficult enough to get decent human transcription from some historical interviews.

    One of my clever archival colleagues had an employee train the software program to recognize her voice. Then she spoke all the Q&A from an oral history interview into a mic as she heard it through the headphones. Voila! Sorta kinda instant transcription. Faster than typing, anyway. :p

  13. Great post.

    Inspirational. Motivational. Humbling.

    Now I’ve got one less reason to not get as much done as I think I should.

    Jon rocks!

  14. I started using Dragon when severe RSI meant I couldn’t type a word. Now that the RSI is thankfully nearly gone, I still use Dragon to write all my articles – it’s so much faster and easier than typing it out.

    It’s not perfect, but much better than I ever expected. I’m pretty finicky when it comes to writing, so if it satisfies me…

    Plus I was also pleasantly surprised by the learning curve – it seemed very quick for me. It’s important to let Dragon scan the documents in your hard drive, so that it picks up your specialist vocabulary.

    Jon was very helpful last year when I e-mailed him for advice about Dragon. And this video is a great resource for anyone considering making the leap to speech. Thanks Jon!

  15. Hi Sonia – sometimes I use Dragon Naturally Speaking speech recognition software if I have a lot to type. It doesn’t take too long to train it to recognise your voice.

    The only downside is that if you have a lot of work to get through – you soon wind up with a really sore throat.

  16. Great article :) thanks

  17. It is interesting note that blogger using ASR software to write article. I think ASR works quite well when it is ‘trained’ properly.
    We use ASR in IVR systems where voice quality is dependent on telephone and mobile network, noise in the background. But in a quite reading room, without back ground noise and efficient MIC, ASR should produce much better results.

  18. It most certainly works for non-English speakers to Dana @ Online Knowledge. I am not native American and my French friend uses it just fine too, not to mention many others. You just have to spend some time training it.

  19. @Eileen: Logitech microphones aren’t bad, but they aren’t particularly good, either. I’d recommend picking up one of the $40 microphones at knowbrainer.com, plus a sound pod, if you’re not already using one. I’d guess you would see a significant increase in accuracy. Also, Dragon NaturallySpeaking works very well in Word. If it’s not working for you, then you probably have a problem with your installation.

    @David: Dragon NaturallySpeaking now supports UK English, as well as regional dialects. And yes, the learning speed is vastly improved. It takes less than 10 minutes.

    @Farnoosh: Yes, Dragon NaturallySpeaking doesn’t work very well with Gmail. It’s one of my biggest gripes about the software. Supposedly, it’s something they are working on though.

    @Dana: Dragon NaturallySpeaking has different speech libraries for non-native speakers, so it should understand you perfectly well, assuming you’re speaking proper English.

    @Cath: Yes, it does wear on your voice! I’m almost always hoarse. Still, it’s a worthwhile trade-off.

  20. Thanks for sharing this information about speech recognition software and how to do it effectively. It’s been something I’ve considered for quite a while, and you have spurred me to get serious about it.

  21. Thanks for the post.
    Like many others I have wondered about this type of software particularly for writing long articles or books. I guess its worth the effort to try something new even though I don’t have the issues of RSI or sore shoulders which seems to have been the impetus for a number of commenters to this blog!

  22. I knew there was a better way to write thousands of articles and do it proficiently! Thank you for that solution.

    I have heard about Dragon but really have not checked into it any further. However, after reading this GREAT post, I’m going right now to do my research and see if it will help me.

    My only concern is that most of the time I’m writing, it comes to me as I’m writing. It seems as though you might have to have an article or speech already created or certainly know what you want to say before this software would be beneficial.

    Any special techniques, tips or advice in that area?

    Thanks SO much!

    Deb Lamb
    Freelance Writer and Marketing Consultant

  23. With this speech recognition you won’t have to think twice before writing and your mind and tongue will work instead of your mind, tongue and hand.

    So this one is sounding great to me … use it and let us know so that we can also shift it too.

    Most important thing is this can save your precious time.

  24. You know…I never thought about using this for my own work because coding themes all day takes it’s toll on my hands and fingers, and then there’s the articles, tutorials, demo site content….etc. I should try this out and see how it goes because if I can help prevent carpal tunnel syndrome, it’s worth the try. Especially that I’m on the computer nothing less than 12 hours a day.

  25. The thing that surprised me when I use speech recognition software, is that I have to *write* differently.

    I didn’t realize how much I think on paper, until I tried to *dictate* to the software. I notice a big difference in my thought process between just saying my thoughts, vs. typing them out.

    I do like it for dumping rapid notes (for example, I’ll take notes during the week on a sticky pad, then dump to notepad with Dragon Speak.)

  26. I’m definitely going to try out Dragon; have heard enough rave reviews about it to want to see what it’s all about.

    I once had pre-installed software on Windows Vista with 07 that worked very well during training but when it came to the real deal, gave me a hard time.

    Still, when I think of all those copious notes some’body’ else can type up for me…

    @ Jon, you never told about how you spoke so effortlessly on that video, without notes. Once more, thank you for your inspirational story and advice

  27. Hi guys

    I didn’t use such voice recognition yet and do not know if this will work out for me but as far as comments and reviews concern, Speech recognition is really something. I want to try it if this will works out for me.

    Kind regards

    Sam
    X

  28. Even after the above post, people are little skeptic about using Dragon naturally speaking.
    I can speak from a Fast writer’s perspective (about 100 wpm). It took me some effort to shift from typing to get going on Speech recognition software.
    My fingers type and follows what my brain spits out. And when you shift into speaking mode, you’ll realize your brain can go higher and you mouth can’t follow and just gets paralyzed.
    It takes some time to get used to it.
    Another thing, you should spend time training your software to recognize your words and speech pattern. The more you talk, the more you use, the better it gets.
    One of the side effect of using this was, I started speaking a lot clearly as I was pronouncing the words correctly for software to understand.
    Varun Pratap

  29. During my college days we have done a project on voice recognition program and made a detailed report about it.On the basic of all that knowledge I can say that this program really works awesome.

  30. @Dana and @Sonia: I’m going to go watch the video momentarily so might get my answer, but, I’m thinking Dragon (and I’d hope any other similar program) had a page of text, a story of sorts, that you read into the microphone. It knows the words you’re reading, and hears how you pronounce them. It then uses that as the starting point for understanding your particular speech patterns and is thereby supposed to give you a good foundation to start with.

    Everything works in theory, right?

    During a demo of Dragon several years ago, I read the story, let it absorb and adjust, then took a few minutes dictating a practice letter. It had very few mistakes. But it’s still starting with a limited dictionary of your voice and so definitely needs tweaking, as previous comments show.

    I did this at the Cleveland State U. Assistive Technology Lab. If anyone is near a university, you might consider checking with them to see if they have such a lab where you can go and get some software demonstrated for you. There’s some neat stuff out there for people with all kinds of impairments to be able to access computers.

  31. i have been thinking about speech recognition software for some time but haven’t taken the plunge.

    i was considering dragon naturally speaking because of a recommendation from a friend. now after read this and seing all the comments it’s time to take the plunge

  32. So, where can I buy this speech recognition software from?

  33. A lot of people dismiss speech recognition software as (still) being less efficient than typing. While this is debatable, it’s sort of missing the point anyway.

    Speech recognition software is best used when you are either (a) attempting to perform two tasks at the same time, or (b) not able to type efficiently in the first place (for whatever reason). It’s not supposed to replace typing. It’s supposed to be different from it.

  34. Speech recognition software should not be seen as a replacement for transcription services. It can be used for short emails and texts, but is not really a viable alternative to professional typing services.