How to get Dragon NaturallySpeaking to dictate into OS x applications

January 11th, 2008 by Kavan Wolfe

iListen is bullshit. There, I said it. The accuracy rate isn’t nearly as good as Dragon NaturallySpeaking. So, I was determined to get Dragon working under OSx, and not just in virtualization. I wanted Dragon to dictate into Camino, Pages, Adium, etc. This how-to explains how I did it. Please bear with me, this is my first how-to.

Warnings

1. If you are a newbie, you may need help for this. A lot of things can go wrong, and this is not the most detailed how-to.
2. I have tested this on exactly 1 hardware setup: mine. I have an Intel-based Macbook connected via ethernet cable to a router. I am using a USB Logitech headset.
3. I admit, the resulting set up is a little buggy.

What you will need

1. OSx Leopard (this may work in other versions, but I have not tried it)
2. VMware Fusion (I use the full version, not the demo)
3. Windows XP installation disk
4. Dragon NaturallySpeaking 9 (Again, may work with other versions)
5. Win2VNC
6. At least one gig of RAM

Part one: installing Dragon

0. Turn off Time Machine (under system preferences –> Time Machine)
1. Install VMware Fusion (this should be pretty simple)
2. Create a Windows XP virtual machine inside VMware fusion, and install Windows XP in it. This should be pretty straightforward
3. When the installation is over, shut down the Windows XP OS, but don’t close VMware
4. go to Virtual Machine-> settings-> memory and set the memory allocated to the virtual machine to at least 512 MB. If you only have one gig of RAM, don’t go any higher.
5. Restart the Windows XP virtual machine
6. Plug in your headset. If it is a USB headset, it will show up at the bottom of the VMware window as a little USB connection symbol. Click on the symbol to assign the headset to the VMware window. When the symbol turns blue, the headset is connected.
7. Put the Dragon CD in the drive. (If you have the CD as an ISO, you can mount it in OS X and then access it from within the virtual machine, as long as it’s shared.)
8. double-click ISScript1050.MSI. you need to install this before you install Dragon. When it’s done,
9. double-click Dragon NaturallySpeaking 9.MSI (it took me a couple tries to install Dragon, as the installation crashed; however, it picked up where it left off and installed without issue the second time.)
10. go through the Dragon set up. If you have audio quality problems, and you have a USB microphone, make sure it’s plugged directly into the computer. Plugging a USB microphone into a USB hub can increase distortion. If you still have audio quality problems, look around the web for pointers. You’re not the first person, and you won’t be the last.
11. You’ll have to read at least one train story (I think) to get started.
12. Once Dragon is configured, open up WordPad and give it a whirl — if it doesn’t work within the virtual machine, it won’t work outside, so make sure you can dictate within VMware before going on.

Part two: dictating into OSx applications

1. In OSx, go to system preferences –> sharing, and turn on screen sharing
2. Click computer settings, and make sure both boxes are checked - don’t forget to enter a VNC password. Leave the screen sharing panel open
3. In VMware, click on the network icon on the bottom right-hand server, and select the bridge option. This connects your virtual machine directly to your home network, and is necessary to get into Win2VNC working.
4. Install Win2VNC.
5. Start Win2VNC, and copy the server info from the OSx Sharing Panel. Only enter the numbers. E.g., mine read vnc://192.168.0.52/, so I entered 192.168.0.52 into Win2VNC. Press enter
6. It should ask for a password. Enter the password you used in the os x sharing panel.
7. If it connects, a little symbol will appear in the task bar. You can change the settings of Win2VNC by right-clicking this symbol.
8. Start Dragon and turn the mic on (the virtual taskbar mic that is)
9. Move your mouse slowly to the right edge of the screen. When the mouse gets to the edge, it will jump to somewhere on the OS X screen (don’t run VMWare in fullscreen mode)
10. Open your favorite OSX editor and start talking.
11. Turn Time Machine back on, if you must.

Known Bugs

1. While using this setup, I find that the Apple VNC Server is very processor intensive, which I suspect indicates some kind of infinite loop going on.
2. Once you’ve started Win2VNC and moved your mouse back to OSx, you’ll have to disconnect the VNC client (by clicking the icon in the OSx menu bar) to get back to the VMWare window.

Conclusion

If you try this, please let me know (in the comments) how it goes. If you figure out what’s causing the above bugs, or how to fix it, by all means share it with the rest of us.

I dictated this entire tutorial into Camino using Dragon NaturallySpeaking configured as described. Good luck.

6 Responses to “How to get Dragon NaturallySpeaking to dictate into OS x applications”

  1. Jonathan Thomas Says:

    Cool tutorial and I almost did it. Good news though, Macspeech, the makers of iListen are ditching it in favor of a port of Dragon, coming out in February. They just announced it at Macworld.

  2. Adam Law Says:

    I was interested to read your account. However, soon you will be able to use Dragon “off the peg” directly on the Mac. MacSpeech are dumping ilisten and have a new product that uses the Dragon voice recognition engine they have somehow obtained from Nuance. Here is the URL

    > http://www.macspeech.com/pages.php?pID=53&osCsid=a9d639ddf7dfdc0b358589df81897dbd

    BTW I have used Dragon Naturally Speaking on my MacPro using Parallels. I read about using a VNC to directly use this in OSX apps, but never tried it.

    ps found your blot because I was reading a book on Service Oriented Architecture by Marks and Bell. It sounded like bullshit. So I googled Service Oriented Architecture bullshit and went straight to your blog. I am a MD so I look forward to reading some of your health reflections. I am now engaged in trying to guide a 100 bed hospital in whether to ditch their heritage system (Meditech) or to try to add a kindly front-end to make it seem useful. Given I have no background in IT and am a busy physician I am not sure I am equal to the task! But distinguishing pearls from bullshit in a subject area that is not your own is a process well worth reflecting upon. I can easily identify bullshit in medicine - but IT….

  3. Kavan Wolfe Says:

    Wouldn’t you know it, I heard about Macspeech Dictate less than a week after writing this…

  4. DB Says:

    Thank you !
    It works very well with parallels tools as well. This is exactly what I was looking for !

  5. Dave Says:

    Kavan

    Been looking for such a workaround for some time and by the sound of the MacSpeech blogs I have encountered your option may well be the more satisfying experience.

    Have you explored any of the Mac command control elements to such a work around (ie. use the DNS to control the mac environment like open aps, navigate pages etc)?

    Cheers for your insights here, Dave

  6. Kavan Wolfe Says:

    @Dave, I haven’t run into any problems, but I haven’t tried much of the OS control stuff. Lately I’ve been using Dictate. It’s essentially an alpha release that they called a full release and sold. The people at MacSpeech should be ashamed of themselves.

Leave a Reply