AI Created This YouTube Video for Me
Testing Steve AI
Steve AI claims to be “The Only AI to Create Live and Animated Videos Using Text.” While the solution does automate the production of some types of videos, the AI element is fairly light.
AI is used to match the text in the script with a library of stock videos. You might consider that more of a search function than AI. Regardless, there is a synthetic voiceover, so the addition of text-to-speech does introduce at least one purely AI element.
How it Works
The solution is pretty simple if you already have a script. Upload the script, and Steve AI attempts to separate the content by scenes. The scenes are generally just sentences that are separated so each segment can use a new image. You have the option of adding a music overlay and whether the imagery is illustration or photos and short video clips. For my demonstration video, I chose the free stock video clips option.
When you create the first draft of the video, you can also swap out the images. I had to do this because the intelligence behind the AI algorithm was not that proficient at expressing the main point of each sentence (i.e. scene). The text-to-video generator also inconveniently sprinkled in some stock videos that required royalty payments. That meant I had to swap out those videos as well to find a free alternative.
The Video Demo of Steve AI
The video was created entirely with Steve AI, except for where a human (i.e., me) was involved. Steve AI asks that you start with some content for the video. The options are to manually upload a script, post a URL of a blog post, or upload an existing audio file (i.e., voiceover). I assume in the audio file scenario, the system is simply transcribing the text to attempt to segment out the scenes and select appropriate images.
Providing the blog post URL didn’t work very well for me. For expedience, I copied the link for the Synthedia article about Canva adding a text-to-image generation feature. The article was apparently longer than Steve AI preferred, so it attempted to summarize the post and reduce the word count to be more appropriate for the script. The summarization made the text hard to follow. Also, I couldn’t figure out how to edit it immediately, so I simply reverted to copying the text from the same post and adding it to the script.
Is the Text-to-Video Good?
What you think about the quality of the Steve AI video will be largely driven by your expectations. Are you expecting something the quality of top YouTube channels? If so, you will be disappointed. Are you going to compare it to product or company explainer videos? In that case, you may be pleasantly surprised.
The entire process took me about an hour. Part of that was because I was learning how to use the solution (which is stunningly easy), editing the script, and replacing about half of the stock video clips with the solutions’ image selector. If I had everything ready, I think this could be done in about 15-20 minutes. Given that experience, it is actually pretty good. The quality is not amazing, but the visual composition is relatively good, and the user has the ability to swap out images and video clips. You cannot create a YouTube-ready video that quickly using traditional methods.
Let me know what you think about the video.
