Speech recognition requires heavy lifting in terms of processing power; heavier lifting than our phones can manage on their own. So how is it that an app like Nuance’s Dragon Dictation, available for free in the App Store, can function on the iPhone? The trick is that the software offloads the processing to Nuance’s servers in the cloud. You speak, the software listens, your words are transmitted to the Nuance servers and the deciphered text is returned to your phone’s screen, all in a matter of seconds. The process works best if you have a wireless connection and demands at least a solid signal from your phone carrier.
The biggest challenge for speech recognition software is that it requires almost 100% accuracy to make it usable. Although a 99% accuracy rate might seem good, that apparently meagre error rate means you have to trawl through everything you dictate to correct the one word in a hundred that slips through. While that’s not really a big deal on a computer, on a mobile phone, where editing is often a painful task, it’s more significant. That makes it vital to ensure you know how to use your speech recognition software properly. Here we'll show you how.
Speech on the iPhone
The iPhone has in-built voice control and apps such as Vlingo provide similar support for voice commands. However, with its limited vocabulary and restricted use, voice control is far more limited than speech recognition.
Google Voice and Dragon Search come closer to full speech recognition, letting you search the web using your voice, but Dragon Dictation is the only fully fledged dictation app for the iPhone. (On Android phones, while you can use Google Voice from most input screens, its relatively wimpy processing power limits the amount of text it can handle in a single bite.)
The neat thing about using the two Dragon apps is that each gets better over time, learning from you as you speak and then correcting its results. Even better, the two apps share the same voice profile, so if you use Dragon Dictation a lot, Dragon Search’s accuracy will improve and vice versa.
Improving speech recognition
Dragon Dictation and Dragon Search provide pretty good results almost immediately, but there are several things you can do to improve accuracy.
- After installing each app, open it, click the Settings button in the lower-right corner and select Language —> ‘English (Australia)’ in order to get the best recognition results.
- Hold your phone about 20cm from your mouth and speak using normal tone, volume and speed. Remember, if you’re using an iPad, the microphone is at the top.
- When using Dragon Dictation, record in small chunks. There’s a 60-second limit on recording, but you’ll probably find using shorter bursts will improve the accuracy. You can continue your dictation simply by tapping the Record button again.
- If you’re going to perform a long dictation session, make sure you use a wired headset with a microphone.
- Avoid noisy environments. The rule is, if it’s too noisy for a normal phone conversation, it’s too noisy for accurate speech recognition.
- Don’t let others use your Dragon apps because it will mess up your voice profile.
- Dragon Dictation is the more accurate of the two apps — due to increased processing power, perhaps — so the more you use it, the more the accuracy of both apps will improve. For example, initially the only way to get Dragon Search to recognise “scones” in the phrase “recipe for pumpkin date scones” was to use the American pronunciation. However, after trying the same sentence in Dragon Dictation, which recognised it immediately, Dragon Search could recognise the Australian pronunciation.
- Correct the app’s mistakes. This is probably the most important step to ensure better accuracy. The more regularly you correct the app, the faster it will adapt to your voice.
Correcting mistakes
You can’t correct mistakes in Dragon Search, but the app does let you choose between different transcriptions. To do so, speak your search phrase and when the results are displayed, tap the down arrow at the right-hand end of the search box to choose between transcriptions. If you don’t see a down arrow, it means Dragon Search came up with a single transcription only.
You’ll get much more flexibility in correcting mistakes in Dragon Dictation. Because of that, if you want to improve the efficiency of both apps, spend a little time in Dragon Dictation, dictating and correcting passages of text. The time you spend doing this will help improve the accuracy of both apps. Here’s how to correct mistakes in Dragon Dictation.
- Dictate some text.
- Tap and hold the incorrect word to display a list of alternatives. If the only option you can see is Delete, it means that there are no available alternatives.
- Tap one of the proffered replacements to replace the incorrect word.
If none of the alternatives are correct, you can get more editing flexibility by using the keyboard.
- Tap the keyboard icon in the lower-left corner of the screen.
- Double-tap an incorrect word and choose Replace from the pop-up options.
- Select one of the suggested replacements or type your own.
Formatting text
Dragon Dictation lets you use key words and phrases to insert special characters and format text. The table on the right shows most of the common symbols you can insert by saying the symbol’s name. If you experiment, you’ll probably come across some more obscure characters to add to this list.
The table on the previous page shows the keywords you can use to alter capitalisation and space your text.
Using your text
The text you dictate isn’t much use unless you can get it out of Dragon Dictation and into something else — an email, an SMS, a notes app, a tweet or a Facebook status update. It’s important to remember that Dragon Dictation functions as a temporary scratchpad, where anything you dictate is stored in the app only during the current session. If you delete it or close the app, your text is lost.
To use your text, tap the Export icon in the bottom-right of the screen. You’ll see an array of destinations, including SMS, email, Facebook and Twitter. Tapping any of these four options will automatically copy all the text you’ve dictated and paste it into the relevant app, ready for sending. Alternatively, you can tap the Copy button to copy the text to the clipboard, then open any app, tap twice and select Paste to paste the text into the app.
