The last time I spent any time with voice-to-text software, it was the 1990s, and the technology was awful. Back then, most programs needed to first learn your voice, a process that involved speaking into your computer’s wired microphone for extended periods of time and hoping that things went well. Between the slow rate of speech required and the high frequency of errors, it never seemed worth the effort to me — especially to a guy who can type nearly 100wpm.
Voice-to-text software has made a comeback, thanks to smartphones. Five years ago voice dialing was all the rage, but today’s users want to be able to tweet, update Facebook, and send e-mail and text messages via voice. Enter Dragon Dictation for the iPhone (and Android).
Upon launching, Dragon Dictation’s interface contains a single button. Press it, and the app will begin recording your voice. When you finish speaking, the app will transfer your captured speech to their servers, convert it to text, and send the text back to you. Assuming the text does not need to be corrected, it can then be copied to your clipboard. For correcting text, users can either re-record words, or choose from a pre-defined list of similar sounding words.
Post-recording, Dragon Dictation provides users with five handy icons: SMS, E-Mail, Facebook, Twitter, and Copy. The Facebook and Twitter icons (once configured) will allow you to send your voice message as a status update. The SMS and E-Mail icons will launch the appropriate app with your message copied to the clipboard — it’s up to you to paste the message into the message body.
When Dragon Dictation works, it works well. When speaking in a normal tone in a normal environment (cars included), Dragon Dictation does a good job of recognizing most common phrases. The program also inherently recognizes certain key phrases (“period”, “comma”, and “Caps”) that allow your text to read naturally.
Using the app while speaking in a lower-than-normal tone of voice and/or in a noisy environment quickly reveals the application’s limitations. So does using non-standard phrases, apparently. My attempt to translate “Ding, fries are done” resulted in “Do you guys are doing”. Attempts to convert speech with the car windows down or while whispering gave similar results. According to the manual, repeated use combined with user corrections will train the app to your voice over time.
Security pundits should take note that all converted text is sent to and from Dragon’s servers via your phone’s Internet connection … meaning members of Al-Qaeda and supporters of WikiLeaks should probably look for a different solution. Dragon Dictation also requests to send the names of all your phone’s contacts to its servers, to ease in voice-to-text translations. The app promises that you cannot be identified and the names of your contacts will never be revealed to anyone else, but the idea still made me a little queasy. Opting out is always an option.
The app contains few other bells and whistles. Users can reset the app’s learned behavior and attempt to detect when you are done speaking, but that’s about it.
Users who have yet to upgrade to a multitasking version of iOS (such as myself) may experience some frustration in using Dragon Dictation for sending text messages. It works, but after each message you’ll have to press SMS, select the person you wish to text, press the text area, press it again, press paste, and then press send. For longer messages it will still save you time, and it’s certainly safer while driving, but it’s far from hands-free. For me, the app seemed more useful for updating Twitter and Facebook, and/or drafting e-mails.
Save for the few stares I got from my family as I clearly spoke Twitter updates into my phone, I like this app. I was able to compose multiple Twitter updates and lengthy e-mails with few or no errors (the more common the words and phrases, the better the success rate). It’s not quite hands-free and it’s not perfect, but it’s not bad, especially while driving.