New AI tools launch almost daily that can transcribe, summarize and re-package your podcast into bite-size promotion nuggets. ClipGen is the latest gladiator in this arena. ClipGen edits moments from your podcast into video clips to share on social media. But, as ClipGen says, “The problem is finding and editing the best 30-90 second clips from a 1-3 hour podcast can be expensive and time-consuming.” So, ClipGen uses an algorithm to find the parts which are “most likely to perform well on social media.” Can ClipGen pick out your greatest hits, summarize and package them for your podcast promotion? Let’s find out.
What Is ClipGen?
ClipGen is a web-based software. You upload the audio file, video file, or provide the YouTube URL. ClipGen then transcribes your podcast episode and uses an algorithm to pick the best parts. You can fine-tune the clips, download and share them on YouTube, Instagram, Threads, or any other social media platform.
I asked co-founder Ollie Crow what ClipGen’s selection criteria is. With videos, ClipGen uses facial recognition and tracking to determine which person is talking, and how. This won’t work for audio-only podcasts. He said ClipGen’s algorithm uses “things like keywords, pauses, laughter, etc.”
“As we’ve gained more and more feedback from customers, we’ve added more features into the scoring algorithm and tweaked thresholds. Now, it scores all possible clips and effectively takes the top 10-12.”
How Much Does ClipGen Cost?
ClipGen starts with a Free tier, no credit card required. After that, the pricing and features are determined by the amount of data, number of clips, and image clarity. (either 720p or 1080p).
Free: Upload up to 60 minutes of audio or video, to make up to five video clips at 720p, per month.
Creator: For £15.00, you can upload up to 120 minutes of audio or video and make up to 20 720p clips per month.
Pro: For £30.00 per month, you can upload up to 120 minutes of audio or video, and make up to 30 clips at 1080p.
The prices are adjustable within each tier. Also, the sliders on their pricing page are fun to play with.
How is ClipGen’s User Experience?
ClipGen’s guidance is minimalist. Some apps offer walkthroughs showing what parts of the screen do different tasks. Others explain what different parts of the screen are if you hover your cursor on or near them. ClipGen’s dashboard is meant for true creatives who eschew rules or structure and click on things until something happens.
The instructions consist of a button that says, “Need Help?” This takes you to a YouTube video tutorial. Text is so 20th century.
But everyone learns differently: some people need printed instructions. ClipGen lacks written instructions, a knowledge base, or support articles. I’m surprised they didn’t have an AI tool transcribe the tutorial video and process it for clarity.
How Does ClipGen Work?
Upload your audio or video file, or enter the link to your YouTube video. ClipGen transcribes and analyzes the file. According to ClipGen’s estimates, a 45-minute podcast episode takes about 20 minutes to process, and a 2-3 hour episode about 30 minutes. You won’t receive a notification when processing is complete: you’ll have to wait and hit refresh.
ClipGen’s clips each have a description, keywords with hashtags, and an editable transcript. Pick the aspect ratio and colors for emphasized words or the waveform at the bottom. The result is similar to an audiogram you could make with Headliner or Wavve.
The user interface assumes you know the numbers of which aspect ratios are best to use on which social media platform. I used a 1:1 ratio. Unlike Headliner, users can’t adjust where the waveform or captions are in the screen layout. Here’s one of the clips that ClipGen selected.
As you can see above, the captions are over the logo, which makes them harder to read.
ClipGen’s Selections and Descriptions
Using AI to choose the “best” or “most interesting” sections feels awkward. It’s like asking a calculator if 7 or 9 is nicer.
In the episode I uploaded, we discussed a deck of cards with writing exercises. ClipGen picked out a section where we talked about a writing prompt, “List three things your character wants. How could each of them lead to ruin?” Then we gave a sample answer that takes place during the Trojan War.
The transcription’s spelling and accuracy were excellent. But, ClipGen focused on the Trojan War and suggested the title, “The Perils of Love, House and Job.” ClipGen’s suggested hashtags were #loveintereststalker, #newhousetrap, and #jobsmurder. How this promotes a podcast clip about writing exercises is beyond me.
AI does a great job transcribing text and summarizing it quickly. But, it doesn’t glean meaning from metaphor the way humans do.
Is ClipGen.io Your Podcast’s Promotion Genie?
The Battle of Yavin sequence in Star Wars is widely regarded as one of the most thrilling cinema sequences ever. When George Lucas’ original editor turned in a rough cut that wasn’t good, Marcia Lucas took over. Not only did she have substantial editing experience, but also she intimately understood the story. Every moment of that intense, heart-pounding battle is due to her painstaking craft as an editor. Because she understood the story and the footage inside and out, watching every frame, she was able to show it at best advantage.
ClipGen’s team says the tool saves you from having to listen to a long podcast. This might be a good tool if you’re a social media manager, making clips for podcasts you dislike or don’t understand. But, if you’re not listening to the podcast to find out what happens and what it means, you run the risk of misrepresenting your client.
Most importantly, if you make a podcast, but can’t be bothered to listen to footage you recorded in order to promote it, why are you making a podcast in the first place?
Our podcast promotion guide has loads of ways to help your show reach new audiences, including making shareable video clips. Can ClipGen pick out sections of your podcast and package them for promotion? Sure. Does it know what will grab attention on social media, or what’s the “best?” That’s too subjective for a machine to find. Should you hand off “interestingness” to a machine? Probably not.
Have you tried out ClipGen on your own podcast? How did you get on? We’d love to hear about it in the IndiePod Community.