Spokesrobots

As discussed last week, AI, if not in it's infancy is certainly in it's awkward teen years. From violations of the laws of physics to stilted, slightly "off" copy, the signs are often there.  This is perhaps evidenced to most effect in the emerging genre of AI spokespeople and video actors.

I gave synthesia.ai a spin and learned that while you can have pretty much any AI-generated character speak in any voice, with any style of narration, getting them to feel real is a bit harder.

Like so many AI platforms synthesia combines AI copy creation, with image search, and other AI tools before then doing its particular party trick: rendering a talking-head video for training, sales, education and a host of other applications.

After selecting one of the templated "characters", I then chose a voice, and a style ("excited!").  Then all it takes is a written script for the app to read. Here are the results. Not bad for a first attempt but it'll be a long time before this appears to be a real actor speaking to camera on the brand or company's behalf:

https://studio.d-id.com/share?id=110c47be65adbaee4d1c88eccfd27d7d&utm_source=copy

To see what else can be done with this tool, I used an illustrative style spokesperson (it's possible to create your own but the process is a little clunky, unreliable and I think would yield some really weird videos. On second thoughts maybe I should try that!). Here's a different take:

https://studio.d-id.com/share?id=c1686066a635491094aabed59e79a048&utm_source=copy

She's right -- you can find out more about AI by checking out any of my recent posts.

#AI #Marketing