To write a review of a speech recognition application, would be a simple task I thought. After all, it all comes down to how whether the application translates to a spoken word into written speech. But, as I went through it, I found that it is a little more complex that taking your vocal sounds and making sense out of them. So without wasting your precious time, let’s go over MacSpeech Dictate.
The Installation
After going to the necessary installation procedure you are required to train your voice. Each voice is stored in a separate profile. The training procedure is ready whether done to the point that once you go through it you almost know how to speak to your computer. You are asked to read out instructions for yourself! It took hardly five minutes after which I felt quite charged up.
A simple as it was for me, my dad had a lot of trouble getting round the first step itself. Not that he didn’t understand the software, but it just wouldn’t recognise his speech. He speaks much clearer than me, and is used to giving speeches to large audiences, so the fault is of the app. It was funny really, when I walked in after ten minutes and found him dearly trying to make his MacBook understand “comma, comma, comma, comma?”.
The Baby Steps
The most important part of speech recognition, is training yourself how to speak to your computer. You do not have to speak in a very robotic voice, but you cannot speak like you would to your friend. During the first week at least, I would get a lot of mistakes in my translation. At that point I would put myself in place of the computer and realized that it’s not the mistake of the application but myself. The app for instance will get confused between “no” and “know”, but intelligently tries to figure out which one to use in context. At least that’s my perception of how it works. Perhaps I just say each word differently in context. At the end of the day speaking a sentence is more likely to get good results than mouthing out a word at a time.
I’ve been using this application for over three weeks now, and I can honestly say that speech-recognition works for me. I have spoken out at least two posts on SA itself, and this one’s going to be the third. I’m sure there are more to come. As for my dad, the speech recognition is slowly working for him, but still nowhere near as good as with me.
Differences in reading and speaking
I’ve noticed that speech recognition is very accurate when you are reading text. That fact that you do not have to concentrate on creating words and just speak them out makes it effortless. Dictating your own thoughts on the other hand, is a totally different trip. At first it is almost impossible to dictate your thoughts and have been translated on the screen. That was one of the reasons why I chose to type out my posts into the speaking them to my computer. But over time, as you get comfortable with your “dictation voice”, it is indeed possible to dictate thoughts.
Navigating your Mac
I found the built-in OS X speech commands extremely accurate even using the built-in mic, so I was super excited about MacSpeech Dictate’s ability to do the same. I’m sad to report, that I am unable to navigate my Mac using my voice. MacSpeech does come with a lot of global commands that can be used to perform tasks such as saving a document, or doing an Exposé of all Windows, but I found these commands to be completely inadequate for moving around your desktop. For instance there is no way for me to directly switch to Mail just by saying it. The latest version of MacSpeech does come with tabbing capabilities so I can switch from the from: field in the subject: line and so on. But sitting on my chair, and surfing the web, while tweeting in EventBox, is not a reality yet.
“Train the words xxx xxx” brings up this window
Navigating text on the other hand, is a breeze. Once you get used to the different commands, like “move backward eight words”, or “copy selection”, editing out a dictated piece is as easy as using a keyboard. In fact, it is advisable not to use your keyboard to edit a document while dictating, as MacSpeech keeps a record of what has been said. Changing anything will get it confused.
The Application Itself
If you’re dictating a lot, the application itself needs to be something pleasant and enjoyable. The MacSpeech UI, baked in native cocoa, is made up of a bunch of floating panels, styled in the default leopard HUD, that include the main control panel, speech commands, recognition windows, and others. You can turn these off at any time and bring them up using just your voice. The control panel features a large button to activate or deactivate MacSpeech. You can also tell it to “go to sleep” and it will ignore any voice input other than “wake up”.
A little too confusing for me, the command editor lets you add new system commands
There are three modes of speech recognition: dictation mode, command mode, and spelling mode. Each of these is self-explanatory and switching between them is effortless using your voice. Overall, while a little buggy at times, MacSpeech feels like a well designed Mac app, in a sense that it is transparent and doesn’t get in the way.
The Perfect Scenario
Like I said, MacSpeech cannot be used in situations where you need total control of the computer using just your voice. However, I am extremely impressed by the applications ability to translate speech. That is what this is all about, isn’t it? If you need to regularly shoot out long e-mails, or write out long articles for your newspaper columns, or if you find yourself regularly typing out your notes from a notebook, MacSpeech Dictate is where it ends. The superior dictation capabilities off this application really amaze me. I taught my Indian accent would get in the way of transcribing, but that was not the case. Perhaps, we just speak a lot more clearly!
On the other hand, if you’re planning on transcribing your college library, reading through all those notes will definitely get you to lose your voice. It’s not easy speaking all day at your computer. Also, writing out a short email, or just a little bit of text here and there is best done with the keyboard. MacSpeech falls somewhere in between the long and short of it all.
At $200, MacSpeech Dictate is not affordable for everybody. First of all, there is no trial for you to download so it’s an all or nothing purchase. The reason for this being that you need a certified noise cancellation headset for it to work properly. It will not even work with a built-in microphone. The box itself comes with a Plantronics headset with other options available on the site. The headset mind you is only good for speaking, as the sound quality of the headset is not worthy for playing music.
I’d suggest you try speaking to your computer first, just out loud and see if you’re comfortable. If you think you can actually use a dictation application, and you are comfortable with the idea of speaking with their computer (it feels a lot weird at first), I don’t think you can go wrong with MacSpeech. If you do decide to purchase MacSpeech, please use our Amazon link and get it at $163.
Screencast
I’ve done a little screencast to show you how it all works out. Note that I haven’t rehearsed any of this to rig the screencast in any way. We’re not affiliated with MacSpeech in any way. As usual, you can watch it here, or download the high quality version using the links below. If you subscribe to our podcast, all our episodes will be automatically downloaded to your system.
If you have any doubts, or if you are a MacSpeech user yourself, please share your thoughts in the comments.
