With over 100,000 apps available for the iPhone, there are a lot of great apps out there. Every once in a while, an app comes along that really impresses me and stands out from the rest. Yesterday, Nuance released one such app, a version of Dragon NaturallySpeaking for the iPhone called Dragon Dictation. Simply dictate a message to your iPhone and, almost instantly, your voice is transcribed with amazing accuracy. With one tap you can send the transcription to an e-mail. You can dictate a message to your iPhone a heck of a lot faster than you can tap a message on the iPhone keyboard, even if you are a good iPhone typist, so with Dragon Dictation you can save a lot of time writing messages or other text. And for a limited time, this amazing app is free. All I can say is, wow.
Well actually, you know I can't help but say a lot more than that one word, so let's get to it. To use the app, simply start it up and tap the red button in the middle of the screen. Then speak your message. You can also say punctuation marks such as "comma", "period", "question mark", "exclamation mark" etc. To start a new paragraph just say "new paragraph".
[UPDATE 12/10/09: Joet of Just Another iPhone Blog has posted a bunch of dictation tips, including instructions on telling the app to capitalize a word (just say "cap" before the word) or to even turn on and off all caps, and including tips on dictating punctuation and symbols.]
You can go for about 20 seconds at a time, and then the app will stop recording and show you what it has so far. Or, you can stop or pause yourself by tapping the Done button. Your speech is quickly translated and then you will see the text on your screen. If you want to add more text, just tap the red recording button again to add to your message.
In my tests, the accuracy of the transcription has been excellent. The app even adapts to your voice over time to improve accuracy, and thus Nuance recommends that you avoid letting other people use the Dragon Dictation app on your iPhone. If this matters to you, note that the app resists translating curse words — an issue of no consequence to me except that I love that it resulted in this amusing post by Andy Ihnatko.
Of course, like all voice-to-text systems, the transcription is it not 100% perfect. Fortunately, there are three easy ways to correct transcription errors. First, you can tap on a word that you want to correct, and a drop-down list will pop up with alternative suggestions. You can also correct an entire phrase by dragging your finger to highlight a few words.
Second, once you have selected the incorrect word, you can tap the red button and record it again. Third, you can tap the keyboard button at the bottom left and manually correct the error by typing it correctly. When you are using the keyboard, you place the cursor in the correct location by double-tapping in a spot.
When you are done with your message, it is very simple to export it, especially if you want to move the text to an e-mail. Just tap the arrow icon at the bottom right of the screen and select "Send to Mail" in the menu that comes up. Within the app itself, an e-mail panel opens with your message included. Address your e-mail, add a subject and click send, and you are done.
Another option is to "Send to Clipboard." This option copies the text so that, within any other app, you can paste the text. For example, you might want to open the Facebook app, a Twitter client, or a word processor such as Documents to Go or Quickoffice and paste the text there. A final option is "Send to Txt Msg" which copies the text to the clipboard and opens up the Messages app. You need to start a new message manually and then paste the text into your text message.
Nuance posted this short video to YouTube that shows the app in action:
Any time a lawyer has someone else write down the lawyer's words, security is an issue to consider. Note that the transcription does not happen within this app itself. Instead, the app sends your voice to a server operated by Nuance, the transcription happens on that server, and then the text is sent back to your iPhone. And that is not all that is sent to Nuance; the app also sends a list of the names in your iPhone's Contacts (but nothing but the names) so that you can say the name of a person who you know and Dragon can do a good job of figuring out the text. For example, I was amazed when this app correctly transcribed "Ernest Svenson" on the first try until I realized that Dragon already knew about that name from my Contacts list. Ben Patterson of Yahoo! Tech asked Nuance about security and received this response from the developer of Dragon Dictation:
Search queries and dictation requests are transcribed by fully automated speech recognition software, without the use of humans. Data is uploaded and collected in order to improve performance for individual users, and to improve the general performance of the system.
All speech recognition requests and associated data are processed in data centers in the U.S. that meet stringent security and privacy standards; these are the same standards that we use for processing private information in other areas of our business.
The Dragon Dictation application does upload a user’s contact names only (no phone numbers, email addresses, etc.), and the data is used only for optimizing name recognition when a user dictates. It maintains its integrity and security.
Speaking more about the app's use of your Contacts list, Michael Thompson, Senior Vice President & General Manager with Nuance Mobile (the guy in the above video), writes:
As you may have experienced already, Dragon Dictation for the iPhone goes through your contact list on your iPhone and uploads the names to our server. We do this for a pretty simple reason: we found that people are often dictating names from their address book and expect the names to be recognized. We take this information and create an anonymous user profile for your device that understands what names are likely to dictate into a document. It's important to note that we only upload the names, not the e-mail addresses, phone numbers or any other personally identifying information from your contacts.
Even though there is no personally identifying information, we still treat all of this information with the highest privacy standards. All of our servers are located in the United States and meet the most stringent privacy and security standards. We conform to these high standards because we use the same data centers for other areas of our business where we are required to store personal information.
[UPDATE 12/10/09: Dragon announced today that it is working on an update to the app that will allow you to turn off the feature of uploading the names of your contacts to Nuance. Thus, if this is a privacy concern for you, there should be a solution soon.] [UPDATE #2 12/10/09: As David Pogue of the New York Times points out, it is a little strange for people to be so worried about Nuance getting your list of contacts when so many other companies have so much more information about you already.] [UPDATE 2/16/09: Version 1.1 of the app is now available, which allows you to choose to not upload Contacts data and to delete any previously uploaded Contacts data from the Nuance server.]
Because the app talks to a Nuance server to do its magic, you need to have a network connection to use the app. On WiFi, I find that transcription is almost instantaneous. Even on 3G, it only takes a few seconds to work. I haven't tried the app on a slower Edge connection.
As amazing as this app is, it is just the beginning. Nuance VP Michael Thompson wrote that the company has "a bunch more applications, languages and features planned." For example, one missing feature is the ability to tell the app to capitalize a word. [See above tip for capitalization.] Also, the app doesn't include a legal or medical dictionary, and I would not be surprised to see this as a future add-on that can be purchased (or a future, separate version of the app).
Nuance is also planning to a launch another app called Dragon Search. They already have a web page set up and the app is in testing now. Dragon Search will allow you to speak your search terms and get simultaneous results from websites such as Google (or Yahoo or Bing), YouTube, Twitter, iTunes and Wikipedia. Google has long offered its free Google Mobile app that allows you to speak your search terms, so presumably the hook for Dragon Search will be the simultaneous results from multiple sites, not just Google.
Dragon Dictate is not the first iPhone app to handle
dictation. For example, Voxie Pro Recorder can record and transcribe text. The app itself costs only $1.99, but you have to pay extra for transcription services, which are done by real people. This is great for accuracy, and they even offer Legal and Medical transcription, but you pay by the word and the cost can get expensive as you can see in these service plans. Another option is QuickVoice2Text Email
which costs only $0.99, but it takes about ten minutes to transcribe your voice. What Dragon has going for it is the years of experience of Dragon
NaturallySpeaking, the premier speech to text engine on the PC and Mac, and the speed and low price of automated transcription by a computer.
As noted above, for a limited time, the Dragon Dictation app is free. Nuance has not announced what the price will be when this limited time offer ends. I strongly encourage you to run, do not walk, to your nearest App Store and download this app right now, while it is free. I guarantee that the time will come when you will find this app handy to have.