The best text-to-speech voices and apps I have found (quickly ;))

The best text-to-speech voices and apps I have found (quickly ;))

I am working on a project for which I need to use some narration which is meant to sound like it’s being spoken by a computer, so I naturally had a quick look at my options.

A while back I used the Adobe Captivate 4 and as far as I could remember at the time I was quite impressed with its text to speech functionality. However, since then I have acquired a new Windows 7 machine, as well as Dragon NaturallySpeaking, and so I was wondering if sticking with Captivate would still be a good idea. Little by little, nevertheless, as I was getting more and more curious about what worse available out there, I started to include some other applications, too. The result is a list of 10 renderings of pretty much the same text performed by five different applications.

Before giving you the samples, may I just say that for me personally it was unusually difficult to get the text-to-speech engine to give me something useful. I would have thought that the Microsoft Narrator would be able to read text from a Word document, but in actual fact it doesn’t; it’s very good at reading menus, though, which didn’t help me one bit. Having just installed Office 2010 on a Windows 7 machine, I had high hopes for the whole process. However, it took quite a bit of scratching my head and reading around, as well as trying out various wacky suggestions available on the Microsoft fora, before becoming aware of the Speak commands which you can add to your Word 2010 ribbon. Why in Excel 2010 there is both a Speak Cells and a Stop Speaking Cells command is anyone’s guess ๐Ÿ˜‰

I also found it quite funky that, within the same application, the US rendering engine sounded quite a bit better than the U.K. one and I also found it occasionally quite difficult to decide whether their male rendering was better or worse than the female one. Anyway, to cut a long story short, here are the files and you can make your own minds up regarding the quality.

This is the text that all the text-to-speech engines had to read. In a couple of cases, though (Neospeech and NaturalReader), because I was testing an online trial version of the real thing, the engine was limited to 200 characters, so its performance is a little bit shorter than the others’.

Dragos’ experiments with the text-to-speech functionalities of Adobe Captivate 4 and Dragon Naturally Speaking have led to an early implementation within an interactive exercise on the importance of mother tongue knowledge, designed and built for the UK National Network for Interpreting project.

  • My favourite rendition was Julie produced by Neospeech. Brilliant, I thought!
  • Then came the Captivate 4 voices Paul and Kate (in my order of preference) – and that was cool, as I find them easiest to use because of Caaptivate’s design: with all the other applications I had to fire up a sound-editing app to capture the sound; with Captivate, you stick the narration text in the slide notes and click on ‘Convert to speech’. Presto, you have a .wav file which you can export or leave in the project. Clever and nice!
  • The next group was pretty close, with performances from NaturalSoftย NaturalReader‘s Graham, Lucy and Kate, as well as Dragon Naturally Speaking‘s Jennifer and Jane.
  • The last one in my books is Microsoft’s own performance: Anna didn’t really impress me, though it wasn’t toooooo bad, I guess ๐Ÿ˜‰

I’m sure there are lots more engines out there and this was by no means meant to be an exhaustive comparison. It was rather my own quick 6-monthly review to see that what I am using at the moment is still good quality ๐Ÿ˜‰