Synthesia https://www.synthesia.io/ is a platform that generates video from text to create product promotion videos, narrated training videos, corporate onboarding and more.
My first reaction was “wow!” and imagined an AI-powered app that would read a short story or training storyboard and create a full animation based on it. So, my expectations were really high, and I was envisioning creating videos that could change the world – or at least generate serviceable learning and training videos, and perhaps even creative work.
Here's a video I made, where I walk through the platform and also play the brief video I made: https://youtu.be/AMJDR67IaXA?si=gx2NXDfmxyATazi1
I misunderstood the capabilities of the platform, but still, when I experimented with it, I was impressed with what could be done. Basically, Synthesia centers itself around a cast of avatars which are based on real actors, and they can be used to narrate the text in training and promotional videos. The images and voices are generated from AI. To deploy the avatars in productive ways, Synthesia has developed templates, which are professionally designed and which have built in some basics of instructional design and marketing. On the instructional design side of things, they are not as rigorous as they could be, and it’s clear that these templates are points of departure, but not the end product.
What I liked most about Synthesia:
· I love the name! It could be a goddess in ancient Greek mythology, especially if you pronounce it Sin-TAY-zee-a.
· Ease of use is a major “plus” – fit for purpose templates reduce the time of content development, and the fact they are modifiable is a huge “plus.”
· Excellent selection of avatars – they are amazing. The voices are nice, too. That said, the platform allows you to represent yourself or any other person who upload their own videos.
· Templates – whether they be for training or product marketing, the templates feature branching scenarios for adaptive learning, corporate training (compliance, etc.), softskill training, product marketing
· Collaborative capabilities: the platform allows multiple collaborators, and in the case of boo-boos, version history for recovery of work
· The platform claims to have the ability to translate into 145 languages. It does not say how well, accurately, or idiomatically such a task would be performed. My personal feeling is, “Don’t hold your breath” and my second thought was “Caution! Never release unchecked and unreviewed from AI into the wild!!!!”
What I liked less about Synthesia:
· The first thing I noticed when I tried out the program was that Synthesia must review the script and if there is anything that aligns with the program’s “trigger” words, the whole project will be shut down. I experienced that myself. I thought it would be fun to see how Synthesia tackled the idea of marketing / promoting a novel, Todos Santos, which is both sci-fi and horror, with some zombie elements along with scary technology and a deranged scientist. Welp. Synthesia said “NO” and would not stomach such project. I get it. A rogue scientist creating zombies is not a universally appealing premise. That said, what happens if you are doing medical training or launching a medical product? Will you be censored?
· I was really disappointed in the voice-over and the awkward phrasings and pronunciations. I don’t know how easily one can train the voice, but it’s important. Since the main area of competitive advantage for this product is the idea that you can use an AI avatar instead of voice talent or actors, this is an important point. I guess it depends on what you want your ultimate level of quality to be.
· A final little quibble is that the learning templates did not have assessments built in, and I would have hoped for multiple choice quizzes at the very least.
Final Thoughts
It is fascinating to see how products are being developed that utilize AI in various products. They test assumptions about how people learn best online, and also encourage engagement.
As in the case of all AI products, there are ethical issues – for example, in recording and training your own avatar, there could be potential for abuse. Where does the new content reside? Is your image now in the Synthesia cloud and not actually owned or controlled by you? Just wondering…
