When computers first reached the level of performance that supported GUIs and not just command lines, voice-to-text applications arrived. Synthesising the human voice from pre-prepared text inputs was a neat trick, and allowed for some interesting interactions.
Then the text needed to be converted into a form where words were phonetically disassembled, and that took time and often human intervention.
Today these tools, as ably represented by Natural Reader, are much more sophisticated.
They can read the text provided in a document, or OCRed from an image, reading it back in numerous voice genders and styles to create quick voiceover or audio presentation.
As some of this functionality comes pre-installed in Office 356, what does Natural Reader do that makes it worth using and licensing?
- Want to try Natural Reader? Check out the website here (opens in new tab)
Plans and pricing
For online, Windows and Mac application users, a free tier exists that limits users to 20 minutes per day using the premium voices.
For online users, unlimited access to the premium voices, mp3 conversion and more file formats requires Premium tier that costs $9.99 per month or $59.88 per year. You can get access to the Plus tier for a total of over 100 voices for $19.99 per month or $110 a year.
Those that use the apps have three tiers to choose from; Personal ($99.50), Professional ($129.50) and Ultimate ($199.50). And, these tiers are one time payments, not subscriptions.
However, if you wish to commercial redistribute the audio generated by the Natural Reader process you will require a commercial license with a $49 monthly subscription ($588 per year billed annually), or you can have multi-users deals that reduce the cost per seat.
They also have some education options for those that allow all students to use for free, incredibly.
Overall, the cost of this service even for commercial use is very reasonable.
The interface for Natural Reader is as simple as one might imagine, since its very much a point and shoot affair.
A panel is available in the middle of the screen into which text can be pasted, drag-and-dropped or you can load documents from the approved file types.
Alternatively, for those using the online model and the Chrome browser, an extension can be installed that allows any text to be highlighted on a web page to be transferred for transcription.
A top bar controls playback, the voice that you want the system to use and the speed of delivery. A menu on the far left provides extra options, including the ability to edit pronunciations.
This functionality is an important feature. How a word is spelt isn’t always a concrete definition of how it is said. For example, the name Menzies it is pronounced in England how it appears i.e. MEN-zeez. But in Scotland, it becomes ‘MING-iss’, as in old English the Z was the now obsolete English letter ⟨ȝ⟩ or yogh.
For those that want some fun with this, and how different speakers of supposedly the same language say things differently, but the popular herb ‘Oregano’ in Natural Reader and then flip between UK English and US English.
The languages available are English (US and UK), Spanish (Spain, Mexico and USA), French (France and Canada), Portuguese (Portugal and Brazil), German, Italian and Swedish.
And, if you want a laugh, it is even possible to provide text in one language and have a speaker of another deliver it.
What’s more useful about this tool, even in the free version, is that it provides an excellent alternative to professional proofreading, since it is much easier to hear a mangled sentence than it is to see the visual mistake in it.
This use might not be the primary function of Natural Reader, but it is something it does very well, and much better than the technology that is unbuilt into Microsoft Word.
As a bonus to the audio generating tools that Natural Reader provides, it also offers a website reading widget that can be attached to any website by its developers.
Having this widget available is a remarkably useful facility for those who browse the internet with a sight impairment. Accessibility is a real issue for those with disabilities and simple to deploy enhancements like these enhance the engagement of these groups with the companies that use them and make it highly likely that they’ll become regular visitors.
When used the text to be spoken is highlighted by the widget, and then each word is also marked as it is spoken. And, the service can use any of the standard 61 voices from 18 different languages.
Natural Reader also made sure that the functionality works with mobile browsed pages equally well.
There is a free version for those who only expect less than 2,000 pages per day to be read by the widget, and monthly subscriptions for those what have 10,000 likely users, or more, per day.
For those interested in WebReader, the details are here (opens in new tab).
Where this application runs into some issues is with names, technical words and the pronunciation of historical texts. But, given that humans all have issues with these precisely the same things, this isn’t hugely surprising.
With access to the pronunciation editor, it is possible to fix these errors in the future, and for those using the system regularly, the accuracy should be high.
Overall, the only significant problem with Natural Reader is that it is a victim of its substantial success.
Therefore, most regular surfers will have heard the voices that they have many times on YouTube videos, taking away some of the ‘natural’ aspects of this solution.
If you are not using it for voiceovers, then this won’t bother you, but for those that are it is easy to find your content lumped in with a large number of small Vloggers who chose not to voice their videos or pay others to do that job.
What Natural Reader needs to offer is some randomise mechanism that can take their best-known voices and alter them subtly without ruining their impact. It would also be nice to hear some regional accents, as they enhance the authenticity of the audio portrayal.
Please don’t misunderstand, Natural Reader does an excellent job, but we suspect it could be even better with some focused development work with the objectives of making the voices even more natural.
Natural Reader is still the gold standard for voice-to-text for many tasks, and that for occasional use that it remains free is remarkably helpful to many.