AI Can Now Create Realistic Songs And Speech – Here’s A Movie Teaser To Prove It!

Introducing the #AISummer2023 series, a continuation of the original #30DayAIChallenge in Spring. As the field of AI evolves, this series delves deeper into the latest advancements in AI tools and capabilities. It represents not just my practical tinkering over the Summer, but a metaphorical ‘summer’ of AI growth and evolution.

As a follow up the AI generated trailer I created for “Algorithmic Overload” (a non-existent movie), I wanted to test a couple of new tools, particularly Suno, which produces some pretty amazing audio (both human speech as well as songs, which I haven’t seen any other tools do so far) as well as Pika Labs, which creates short 3sec video clips similar to RunwayML.

So I continued where the last trailer left off…the premise that cats will save the world (naturally!). Here’s the final product:

Both the intro narration and the Rap song in this video
was created using AI tools

How I Did It

Here are the tools and techniques used:

• Scriptwriting – As usual, the ideation and scriptwriting (i.e. the narration at the beginning of the teaser and the rap song) was done using ChatGPT.

🎵 “Yo, we got cats in hats, ain’t no time for chitchats, Battling AI, with their ninja combat stats, Swinging their claws, breaking digital laws, Fighting for a cause, no time for a pause, Saving the world, furballs all unfurled, These cats are heroes, give a cheer, everybody that hears!” 🎵

• Visuals – The base visuals were all created on MidJourney and I then took these and used Pika Labs (also available via Discord and currently free, compared to RunwayML). I initially tried video generation using text but the results were pretty poor, so stuck with the image to video approach.
• Sound – I used Suno for the first time and was pretty impressed with its ability to generate realistic sounding speech…and especially songs. Suno is also currently available via Discord and has a pretty simple prompting approach, even allowing for lyrics to be generated directly via ChatGPT. Currently the tool doesn’t offer much control over the generated styles (instead choosing a random genre/style) but I suspect this will change with further testing and development.
• Editing – I used my go-to video editor, Final Cut Pro X…though pretty much any video editing tool would work equally well.

That’s it! Looks like RunwayML and Elevenlabs are going to be seeing some competition from the likes of Pika Labs and Suno. Keen to see how these services evolve in the coming months!