Opinion: Text to Speech instead of Human Narration?

I've been playing with text to speech technology for a while, hoping that one day we can cut out a significant development expense.

Now that SL360 has T2S it as a built in option, it's back on my radar. 

So far the feedback from people in my group about using "robot" narrators is mixed. 

On one hand, we can increase output and narrators provide key engagement at certain points in eLearning coursework (sometimes our budget constraints would not allow us to pay for voice talent at all.)

On the other hand, these voices are far from perfect and still can't exhibit much of the inflection and warmness created by a human voice.

What is your opinion on using text to speech voices as primary narrators?

9 Replies
Matthew Bibby

I prefer human voices, however, TTS is very handy during development to use as a placeholder while waiting for the final audio.

That being said, I have used TTS as a primary narrator before. I sprinkled robot jokes throughout and had the TTS try and say tongue twisters, imitate the "fitter happier more productive" voice from that radiohead song etc. The audience loved it and the use of humour seemed to make the TTS more bearable (and it made sense for the audience and topic). 

Sean Speake

As a learner I detest TTS - for the most part it's pretty terrible.

That being said, using an approach like Matthew's excellent suggestion can be helpful for the right topics.

I'd also strongly suggest keeping it to short elearning bursts. There's a world of difference between a quick hit 5 minute blast and a half hour long course of robot droning.


"Warm and Natural" are words that resonate with me. I might even prefer the "conversational SME speaking passionately, if imperfectly" over the "well rehearsed voice talent".  Most of my participants are used to conference call audio and presentations so hearing the same type of audio in a course seems natural to them.

Governance People

I'm experimenting with Amazon Polly.  People are getting used to Alexa and this is the same technology.  I will not use as yet in my courses, but maybe we are not too far away from it.  You access it through AWS and use specific text markup to alter the speech.  It is not expensive by any means and you have instant access to the mp3 files.

E.J. Smith

Hi Gregory, 

I know that this is an older post, but I figure that I'll weigh in just in case. I'm biased about natural voice and TTS voices because I'm in the voice over industry. My company GM Voices http://www.gmvoices.com/docs/ej/e-learning_rescue/ is behind a large majority of TTS voice options that are on the market today, but with that being said, your learners need as few distractions as possible when they are engaged with your learning modules.

Having a TTS (text to speech) can become a huge distraction if the learner has a difficult time getting past the lack of empathy or emotion of the subject matter. And because not all topics in eLearning are the same, having a natural voice actor as your go to, allows the learner to truly "feel" the emotion, empathy, or severity and eagerness in the tone of the voice actor. 

If you're looking for cost-effective solutions for voice, please let me know how I can help. 

E.J. Smith