Descript Overview & Content Creation Ideas
An AI that isn't [redacted] + a fun "grey" technique
I took some time to check out Descript this week and the TL;DR for a lot of you is this is a tool you should take a serious look at, especially if you’re a Substack writer, SEO person who needs video content, do a lot of screen recording for work, record educational/teaching content, have a podcast, or just need to crank out a lot of social media clips.
The core value prop is you can edit audio and video content with a text based workflow, but it has a bunch of other helpful AI-based features bundled together.
Is it changing the world, replacing jobs, or sending Adobe to the stone ages? Unlikely.
But for certain tasks it can cut down on editing time in a big way. This is an awesome use of AI to make work more efficient.
BowTiedTalon makes a very good point here:
I use a hardware controller with a jog wheel to quickly scroll around a recording when I edit, but being able to just read along is even faster. Being able to search for and edit out “ums” and “uhs” en masse is also a big benefit—this is a real pain using a timeline based editor. Descript also has an automated function to highlight and remove your stuttering and stammering.
I’d actually love a native plugin for Adobe or Resolve that I could drop on to a clip and automatically cut clips that are likely filler words, then be able to delete or adjust as I go.
But for now there’s actually an Adobe integration, so this is something I might use or have an assistant use to prep future projects for me.
Quick Note On Earning Money With This
Something to keep in mind is the BowTied community is months to years ahead on a staggering number of things. If you’re looking for ways to make money and add value with media skills, don’t read this and convince yourself not to bother because “now everyone has this”.
Yes this a tool that is available to whoever, but a lot of people don’t even try or just don’t have the time to deal with it themselves. I will certainly use this tool at some point to fix bad audio and look like a magician.
Then there are the economic problems going on that you can read about over on BowTiedBull which plenty of people will use as an excuse to not even try.
Stuff like this gives you an easy boost. Don’t blow your lead.
The best way to get a feel for a tool like this is to jump in and start doing stuff. And their docs are pretty clear so you shouldn’t have much trouble getting up and running.
No sense in regurgitating their docs and pitch deck for you, but here are examples of things I’d try to play with first, along with some ideas for how to apply them.
Transcribing A Video
You can literally drag and drop in a video or audio, wait a bit, then it will have transcribed your video.
Are you the type of creator that makes YouTube videos from a script of loose bullet points and improvising as you go?
This feature is sick if you want to transcribe the video quickly and get it into your keyword tool like Surfer SEO to adapt it to a blog article and get some extra site traffic. Don’t forget to embed your video on the page when you’re done. Feed the spider.
Remove Filler Words & Add Captions
I touched on this earlier, super helpful.
Even if you plan on editing in a more standard editor like Premiere or Resolve, if you’re cleaning up an hour long podcast episode it makes all kinds of sense to scrub out the fluff inside Descript first, then export to your primary editor.
You also have a couple options with captioning video. You can use their auto captioning feature which offers some basic options.
Or you can use the Descript transcription and dump that into Premiere or even After Effects if you want to have more control over placement and transitions. This can be done manually or you can use Descript to import .srt caption files into Premiere.
It Has A Basic Timeline Editor
Obviously not as robust as a more mature DAW like Logic Pro or Adobe Audition, but you can do some basic tasks right in Descript if you find it easier to look at waveforms, want to automate some crossfades, or need to apply basic effects like compression and EQ.
Create a Model of Your Voice
For the readers who are public or bowtieds who are comfortable with putting their voice out there, you can create a model of your own voice.
Instead of using a canned text-to-speech AI voice, you now can type to talk in your own voice.
If you do a lot of audio content or make videos with voiceover, the obvious benefit here is you can save a ton of time because you don’t have to actually sit and record it, or dub in sections to correct sections where you made a mistake.
Even if you are still going to do live recorded audio, you might want to try folding in some AI audio with the real audio where it makes sense.
Clean Up Your Voice
One of the more magical features of Descript is the Studio Sound AI.
Most people are familiar with basic noise cancellation features and maybe know how to EQ their voice. Or use a de-reverb plugin for room noise if you want to get really fancy.
But Descript does a killer job at making all this very easy and going past what is possible with traditional tools.
You can talk into a crappy phone or computer mic and significantly enhance the sound quality. Recording with a good mic still sounds better for now, but probably not to the level where most people will care.
Kind of similar to how iPhone cameras aren’t going to replace full frame Canon or Sony cameras, but they’re still close enough for the average person.
Hear it for yourself:
Don’t Want To Dox Your Voice & Want To Sound Like A Celebrity Instead? How to clone whatever voice you want.
This is tedious, unlikely to have a long shelf life, and certainly not something they want you doing.
So consider this a “grey hat” technique.
Don’t do this on your work account and do not break the law.
Disclaimer out of the way, this is what to do..