After deciding to pause my attempts to do a complete 'one prompt' script-to-video AI generated short film for kids, I continued looking at alternative workarounds for making the story come to life, by focusing on the individual production elements.
So I started searching around for tools to create AI generated voiceover narration and character voices.
It seems that text-to-voice gen AI has been RAPIDLY evolving while everyone has been busy looking at text-to-video.
I found a number of text-to-voice generators such as:
ElevenLabs (which was my preferred choice and which I'll cover in more detail in my next post)
https://www.topmediai.com/app/text-to-speech/
that pretty accurately generates narration using the voices of Morgan Freeman, Donald Trump, Snoop Dogg and others
and
https://www.tryparrotai.com/ which does the same but ALSO generates short, 'shallowfake' videos of your chosen celebrities saying whatever you want them to say with a text prompt, albeit with some pretty bad lip sync.
While I was on this last site, I decided to take the opportunity to do a little experiment to generate some free celebrity endorsement for Matthew Donaldson Workshops.
Even though doing this raised some serious questions for me about how governments, industry, and social media platforms would be handling the very real potential for malicious misuse of this kind of technology, and what this would mean for the value of video evidence and testimony in the future, I could feel myself being dragged down a very deep rabbit hole!
But now that I had the endorsement of some high-profile celebrities 😂 I felt motivated to re-focus and carry on with my research into how to make character voices and voiceover narration for my own Gen AI short film.
Next post: Elevenlabs.ai for Voiceover and character voice generation
After deciding to pause my attempts to do a complete 'one prompt' script-to-video AI generated short film for kids, I continued looking at alternative workarounds for making the story come to life, by focusing on the individual production elements.
So I started searching around for tools to create AI generated voiceover narration and character voices.
It seems that text-to-voice gen AI has been RAPIDLY evolving while everyone has been busy looking at text-to-video.
I found a number of text-to-voice generators such as:
ElevenLabs (which was my preferred choice and which I'll cover in more detail in my next post)
https://www.topmediai.com/app/text-to-speech/
that pretty accurately generates narration using the voices of Morgan Freeman, Donald Trump, Snoop Dogg and others
and
https://www.tryparrotai.com/ which does the same but ALSO generates short, 'shallowfake' videos of your chosen celebrities saying whatever you want them to say with a text prompt, albeit with some pretty bad lip sync.
While I was on this last site, I decided to take the opportunity to do a little experiment to generate some free celebrity endorsement for my Matthew Donaldson Workshops business.
Even though doing this raised some serious questions for me about how governments, industry, and social media platforms would be handling the very real potential for malicious misuse of this kind of technology, and what this would mean for the value of video evidence and testimony in the future, I could feel myself being dragged down a very deep rabbit hole!
But now that I had the endorsement of some high-profile celebrities 😂 I felt motivated to re-focus and carry on with my research into how to make character voices and voiceover narration for my own Gen AI short film.
Next post: Elevenlabs.ai for Voiceover and character voice generation
In pursuit of the idea of copying and pasting the original short film script about Dinosaurs vs Aliens that ChatGPT & OpenAI had written for me (with lots of flowery language), into some sort of magical AI video making software, I googled around to see if there was some lo-fi version of OpenAI's Sora AI Video that I could use to experiment with.
In a hopeful turn, I found invideo, and started looking more closely into what it did. With little bit of charming, optimistic naivety I thought that I could just copy my script into its text prompt area, and my Gen AI short film masterpiece would be pumped out the other end within minutes.
Disappointingly, but hilariously, this was not the case.
It seems that Invideo trawls through all stock footage available within its dataset, and ASSEMBLES existing stock footage clips rather than GENERATING new clips based on a script to make a short movie.
This was not what I was looking for.
(I did however note it as a useful way to find archive footage for documentaries without time intensive archive searching across multiple libraries.)
For some reason, Invideo also decided to make the already-over-the-top flowery language in the script that I'd given it, feel like it was trying really hard to get itself beaten up by high school bullies.
(Did it want to show that it was smarter than me? Did it want to impress me? Why all of a sudden was I worried that someone was going to try to take its lunch money?)
What did impress me though, was that it generated its very own voiceover based on the script. Which got me curious about what text to narration AI tools were out there.
I decided to pause my text-to-video experiment for the moment, and go looking for other Gen AI tools to start building the individual audio-visual components of my story, while continuing my search for text-to-video workarounds.
In my next post we'll look at text-to-voice for the purposes of simplifying voiceover narration and character voicing, but for the mo
The Importance of Māori Culture in Ending a 10 Year Civil War
Peace Not Guns for Bougainville
Women and Children Affected by War
Guns on Bougainville during World War 2
The Copper Mine That Sparked A 10 Year Civil War
Women and Children Affected By War
The Importance of Land to the Bougainvilleans
Bougainville - An Island Paradise