For some odd reason Nuance decided to drop Dragon Dictate 5 for Mac on a saturday. Anyways here’s a brief first look at the new speech recognition version for mac. Installation.
Dragon Dictate 5 for Mac is being released Globally on September however being a previous customer I was given early access. Be warned that if you purchase from a USA store to try and take advantage of exchange rates you’ll end up paying 20% on top of the price.
Before you can install you need to download an installer to download the installer. The end download comes in at 2.5 gb and takes up 2.8gb of space.
For some reason you’ll need to have Safari closed to do the install. Seems that the installer has a slight problem as well. Being an existing customer I selected “existing customer” but it still wanted me to sign up as a new customer (sigh)
Dragon Dictate 5 First Run
Good news is that existing profiles can be migrated from 4 to 5. Using a 2009 Macbbook pro with a SSD this took an eternity getting seemingly stuck on Activating Grammers. There’s a handy reveal profile in finder so you can do a backup (recommended) before attempting the migration.
After a force quit I tried once again to upgrade my profile but after 15 minutes of waiting once again I was stuck at the activating grammars section. Time to bite the bullet and do another profile from scratch.
The bad news is that with migrating a profile it means you won’t be able to use Dragon Remote microphone, which explains them missing from the migrate profile box.
Creating a New Speech Recognition Profile
The usual settings are there, name a profile, select a microphone and then read a passage of text so that Dragon calibrates to your environment. There’s a good amount of guiding and hinting to say speak like this and you’ll get better results. Training the user is just as important as training the software.
Dragon Dictate 5 contains more microphone profiles than before including ones for built in microphones. As always it’s best to do anything like this in the confines of a quite room for best accuracy.
The usual settings are there, name a profile, select a microphone on either read a passage of text I wait so that DragonDictate Fifa marking calibrate itself to its environment.
The above was dictated without any training so the results are a little inaccurate. Training text is the same as in version 4, limited to 4 training excersises. All of them take about 5 - 10 minutes and you’ll have to do it for each and every microphone or device that you’ll using with Dragon Dictate 5. Which is why I’m miffed that my profiles didn’t migrate across properly.
The results after the 1st training exercise.
The usual settings are there, name a profile, select a microphone and then read a passage of text allowed Dragon calibrates to your environment. There’s a good amount of guiding and hinting to say speak like this and you’ll get better results. Training the user is just as important to training the software.
One thing I don't understand is that with transcription it only takes 60 seconds yet for dictation it's the best part of 40 minutes for training?
So after just 5 minutes of training and carefully reading the text in the same manner as before, as you can see the success rate has vastly improved. I'm begrudgingly doing more training later as my profiles not migrating over has left me somewhat irked.
Unlike Siri, text tends to come in blocks not on a word by word basis that we've become accustomed to like Google or Cortana as well. Also the speed of which text appears on the screen is a touch app dependant. In text edit text snapped onto the screen, using Safari + wordpress there's a definite lag but not that bothersome, just noticeable.
This could be down to the 2.26 GHz Intel Core 2 Duo processor so I'll re run a few tests on a more powerful mac later in the week,
The new look and feel is obviously based on Yosemite and the upcoming El-Cap release. There's no doc icon instead all features are accused via the Dragon icon up in the menu bar where you can access all of Dragon 5's preferences which are a little sparse.
Dealing With Recognition Mistakes
To start with the experience of using voice recognition can be equally frustrating as it is rewarding. Into you have completed all of the training and trained additional words for the vocabulary you will find yourself rereading what you have just spoken many times to pick up any errors.
Dealing with errors can be as simple as saying "choose one" from the correction window. However unlike the previous Windows corrections are shown immediately instead you have to say "show correction window" and then choose the correct entry.
Dragon Dictate for Mac 5 seems happier when you use text edit rather than safari.
Many times I've seen the cursor randomly jump to different places on the screen and end up deleting or replacing text arbitrarily.
Not to mention the incessant nagging about installing the Safari plug-in but at least in this version the tick box is already there to say stop nagging me.
Recapping the first hour.
- Profiles failing to migrate is a huge annoyance. Don't forget every device you have will need to be trained. Thats 3 devices for me so 3 lots of 40 minute training sessions
- UI has a nice look and feel although being very grabby with windows. I.E if a dialogue box is up and you want to drag a window past
- Why on earth doesn't the correction window stay up when I ask it to.
- Function key's for preferences don't work. at least in OS X 10.10.5
Things left to test
- Wheres the darn recognition window gone?
- Looking forward to trying the transcription engine.
- Why do I have to say "show correction window" every time?
- Figure out why preference keys don't work.