Dragon brings speech to the iPhone with Search and Dictation apps

Nuance's Dragon speech recognition tools are well known on PC, but in the mobile world the company's probably best known for its work on extended T9. That's going to change with the arrival in the UK App Store of iPhone and iPad versions of its voice-powered Search and Dictation apps. Initially available for free from the App Store, they're tools that let you use speech recognition anywhere, anytime.

We've been playing with early release versions of the two apps on an iPhone 4 for a couple of days now, getting a good look at what should be in the App Store in a week or two (depending on the gatekeepers at Apple). These aren't the first voice applications for mobile – we've used the speech recognition tools in what's now Microsoft's Bing mobile application on Windows mobile, along with Vlingo on BlackBerry and Google's voice-enabled search tools on iPhone and Android. What Nuance's apps do (like Vlingo's) is put a desktop quality speech recognition engine in the cloud, using your 3G or WiFi connection to send voice to the cloud, and recognised text back to the phone.

Both apps have a very simple user interface – a silver and red button in the middle of the screen titled "Tap and speak". And that's really all you need to do, tap the button and speak in to the phone. A settings icon on the main screen lets you configure the application, setting the language being recognised and in the case of the search application choosing a default search engine (either Google or Yahoo! – there's no support for alternative search engines like Bing).

Once you've had some speech recognised you're dropped into Search's results view. Here you can slide between your chosen search engine and the other information sources Nuance offers. These include Wikipedia, Twitter search, YouTube and iTunes, and are simply web-page views wrapped in the Search user interface. That's not a bad thing, after all, the sites Nuance is using are already formatting their pages for mobile devices, and there really isn't any mileage in reinventing that wheel. There seemed to be an issue with iTunes integration in our pre-release version, and we weren't able to get any results – even when we searched for well-known artists we knew were represented in the iTunes store.

While everything worked well, once we'd got the hang of talking to the phone (just use it like a normal phone!) , we did have one annoyance with the Search app: one you've started searching there's no way to get back to the settings page, which is only available from the initial search button. That means there's no way to quickly switch between search engines.

Dragon Dictation wasn't quite as successful as Search at recognising my speech – though it still did a lot better than other phone-based tools we've used. It did recognise most of what we'd said first time, and tapping on a word gave us a sampling of alternate choices that often contained what we'd actually said – and failing that, there's a keyboard for manual editing. There's no formatting, though Dictation does recognise some punctuation terms, like "exclamation mark", "comma" and "full stop". If you need to delete a section of text, just drag to select and then choose the delete option.

Once you've got your text ready, you can send it by email or SMS, or copy and paste into other applications, or even deliver straight to Facebook or Twitter. As the choice of services implies, this isn't a tool for dictating long diatribes and we'd be more likely to use the desktop alternatives for any serious dictation. However, if all you want is a quick reply to an email, or an update to a social network, then Dictation is the tool for you – especially if you struggle with the iPhone's capacitive keyboard.

So does speech recognition work on the iPhone? The answer's a qualified yes, as we weren't able to test Nuance's tools with a wide selection of regional accents. Out review software was able to work well even with significant amounts of ambient noise, though it did have some issues (like any speech recognition software) with specialist terms and non-English proper nouns. Using these tools is certainly a very different, yet really rather natural way, of working with a smartphone. There's no need to tap at touch keyboards, you just talk. All you need to do is get over the embarrassment of talking to a computer in the middle of the street…

Simon Bisson