oice recognition is becoming vastly popular with its presence in our homes and workplaces increasing each year.
But with that popularity comes an increase in service providers and with that comes the question: which voice recognition system is the best? Well, there is only one way to find out… FIGHTTTTTT!
Just kidding. Instead, we simply decided to test the top six voice recognition engines out there against each other.
Now we know what you’re thinking… We’re a speech recognition provider ourselves, so will it be fair? Our answer to that is yes!
We may be advocates for Dragon speech recognition, but first and foremost we are The Speech Recognition People.
We are passionate about voice recognition software and the impact it can have on people’s personal and professional lives. What’s more, we know the technology better than most and understand the demands of its users.
So, with that said we promise to be fair and non-biased and simply leave you with our findings.
More specifically, we’ll be reading the same passage of text consisting of roughly 260 words. Whittling the criteria down to three aspects, we’ll be testing the solutions under:
- Price – Because no one wants to break the bank!
- Accuracy: How many errors, be it grammar, punctuation or completely the wrong word
- Functionality – Is it easy to use?
Let’s get to it!
Dictation io
Price: Free
Accuracy: 9 errors. Specifically ranging from not capitalising after a full stop and then capitalising when not necessary. Also, inserted unnecessary spacing.
Functionality: Easy to use. A slight inconvenience in having to copy and paste all dictation from the engine to the desired application i.e. Microsoft Word or email. Web-based nature highlights concerns of GDPR and data security as it is not secure.
Our first voice recognition engine is called Dictation io and is entirely web-based. In short, it runs from Google Chrome, Microsoft Edge and other internet browsers.
As shown in the image above, the solution allows you to dictate in a notepad style. Users can then either copy, save, publish, tweet, playback, email, print or clear their dictation.
This is a slight disadvantage as users will have to transfer their dictation to either email, Microsoft or any other application. However, considering it’s web-based and free – not bad!
Google Docs Voice Typing
Price: Free
Accuracy: 6 errors. Common errors include inaccurate hearing and misunderstanding, missing out full-stops and inaccurate capitalisation.
Functionality: Allows free-flowing dictation and is self-explanatory to use. Not compliant with GDPR as it is web-based.
This system is also web-based and created by the tech giant, Google. Similar to Microsoft Word, Google Docs has a detailed taskbar within which Voice Typing is located.
I liked that Google Docs Voice Typing allows free-flowing dictation (something that Dictation io was missing). Users can dictate freely into the document and save/send accordingly.
Apple Dictation
Price: Free (available only if you have an Apple device)
Accuracy: 11 errors ranging from misunderstanding words/inaccurate hearing, missing out punctuation and not capitalising when it should.
Functionality: You can dictate into almost any Apple application i.e. iMessage, WhatsApp, Facebook and Outlook which is handy. Not GDPR compliant. Struggles to ‘listen’ for short periods.
The freedom of dictation across the board is Apple’s main advantage. You can dictation into almost any application, whether you’d like to send a text to your Mum or an email to an important business client, you can do exactly that.
However, this voice recognition engine had a small attention span. While dictating the 260-word passage, at numerous points, the device stopped listening; meaning I had to press the dictation button on the keypad to continue dictating each time.
I can imagine this would be slightly irritating for users wishing to dictate a large report or contract. Furthermore, Apple dictation wasn’t very good at using context.
Windows 10
Price: Free (if you have a Windows on your device)
Accuracy: 21 errors (yikes!). Mainly consisting of incorrect words and incorrect capitalisation.
Functionality: Sorry Microsoft, but this feature lacks functionality. This engine operates via a dictation box even when you’re within the app you’d like to dictate into.
More specifically, I wanted to dictate into Word, however, I had to dictate into the box and say “insert” to put the text onto the document.
What’s worse is that you have to say this command as frequently as after every three sentences.
This was a problem as I attempted to dictate the whole 260-word passage to look up and see that it only recognised the first three sentences as I hadn’t told it to “insert” my dictation.
In conclusion, the high number of errors and the complicated use, Windows 10 voice recognition didn’t live up to my expectations.
Dragon Professional 16
Price: £566.66 +VAT
Accuracy: 2 errors. Soley misinterpretation i.e. it mistook ‘edition’ for ‘addition’ and printed ‘A’ instead of ‘An’.
Functionality: Simple to use with voice commands available to speed up the dictation process, for example, microphone on and scratch that. Correcting errors with Dragon is also easy.
Dragon Professional 16 by Nuance Communications comes in at 1st place with regards to the number of errors. In this case, one of the errors can be easily fixed with some user training.
It is a matter of teaching Dragon: when I say ‘addition’ I want you to print ‘edition’.
Another benefit is the Dragon Voice Commands. They allow you to carry out specific actions via your voice.
Most importantly, corrections like this (and any others) will be remembered instantly; meaning Dragon will not make the same mistake more than once.
It allows you to dictate freely, however, the dictation box feature is available. Dragon is also compatible with EPR systems such as SystmOne and EMIS as well as other case management systems such as DPS software.
Android
Price: Free (available only if you have an Android device).
Accuracy: 10 errors. This speech-to-text engine failed to print some punctuation, for example, it wrote ‘hyphen’ instead of the appropriate symbol: ‘-‘.
There were also missing capital letters and the beginning of sentences.
Functionality: Relatively easy to use, activated by the dictation icon on the keyboard. Unlike the iPhone, Android speech recognition actively listens until you manually stop it.
The Android voice recognition feature comes above Siri in this study, beating it with just one less error.
Similarly, Android has better functionality; it doesn’t have the ‘short listening time’ that I found with Apple. However, both Android and Apple do share one thing in common: they aren’t secure or compliant with GDPR.
So although it is convenient to dictate with your mobile phone, unless you want Big Brother to know the ins and outs of your life, I’d avoid it if you can!
The Results
So, we’ve shown you our findings, but what does that mean for the speech recognition engines comparatively?
Let’s take a look at how they stand in terms of accuracy…
2nd: Google Docs Voice Typing
3rd: Dictation io
4th: Android
5th: Siri
6th: Windows 10
Hopefully, our study has put you on the right path to finding the perfect voice recognition solution for you.
Whichever solution you decide on, make sure you maximise your return on investment by polishing up your dictation skills with our tips.
If you still haven’t made your mind up about the solution you’d like, then get in touch. Our speech technology specialists are happy to have a no-obligation chat.
