James Rosewell writes:
Mark’s been encouraging me to write an opinion piece on the Nexus One for the last few days and I’m finally putting fingers to keyboard to share my experiences. It’s taken so long because this phone has so many features. On a positive note I could go into details about the gorgeous screen, the Android Marketplace that will out-sell Apple’s over the next 18 months, the built-in satellite navigation service and the speedy processor that makes everything run smoothly in real time. Or on a less positive note, the touch screen keyboard that sucks (think carefully about this if you’re a heavy texter or emailer, it’s even worse than the original iPhone), the lack of ActiveSync for Calendars and Tasks, no support for WMA music files or the clunky zoom functions on the web browser.
However I’m going to focus on voice dictation. Nexus One is the first phone I’ve used with this feature.
Voice dictation can be activated by opening the “Android Keyboard Settings” and pressing “Voice Input”. Once enabled, a message appears informing “Voice input is an experimental feature using Google’s networked speech recognition...”. A clear warning this isn’t ready for primetime. The keyboard now has a little microphone button next to the space bar, or a finger can be swiped across the keyboard to activate recording.
When activated the phone starts listening. During a period of silence it assumes the speaker has finished and gets to work turning voice into characters and words. Processing takes about 2 seconds for every 1 second of speech. I imagine this partly depends on the network bandwidth available as the voice recording is translated somewhere in the Google cloud. Once complete the text box will contain Google’s output which can be accepted by typing another character or wiped in its entirety using the delete button.
All well and good. Is it any good, you’re asking? Well the short answer is that it won’t replace the keyboard... yet... and you’re not going to be able to safely send text messages while driving. Any sentence containing names or more than 5 or 6 words will be a disappointment. A few words spoken clearly such as “Hello Nexus One” or “I’m running late” work really well. A more complex message is going to disappoint.
However it’s a start. I’m reminded of my Nokia 2110, the first GSM phone I had with text messaging. Messaging was buried in the menu system, there was no predictive text, moving between upper and lower case was a mess and 160 characters really were the maximum size of a message. When you finally did work out how to send a message, the person you were sending it too couldn’t work out how to respond.
People and technology have moved on a long way since the giddy days of the mid 90s. The same will be true for voice dictation on mobile phones. Dragon Naturally Speaking from Nuance, which recently purchased SpinVox, is a great product for the desktop computer. It requires hours of training to your voice by reading books to it, needs as many CPUs and GHz as you can afford, but it does a pretty good job once you’ve been able to get your head around how to dictate (an art that has been lost following the demise of the typing pool). Given a few years, voice dictation will be standard on all smartphones with network operators and handset manufacturers owning data centres full of voice dictation servers. When that happens, phone user manuals will swell to contain excepts of Moby Dick and other classics to read to the phone during the setup stage.