Stop paying for video editors — do this instead
By Hasan Aboul Hasan
Summary
Topics Covered
- AssemblyAI beats Whisper on tone understanding
- Visuals are generated, not screen-recorded
- Voice quality drives retention more than cuts
- Each video makes the next one cheaper
- Full edit costs one-tenth of a human editor
Full Transcript
Today, I will show you how I made Claude, edit my full YouTube videos and to end from a raw recording till the upload with thumbnails, titles,
everything. And this video you are
everything. And this video you are watching right now is a real example.
Everything you see here, the animations, motion graphics, even the screen recordings, the music, the sound effects, and the thumbnail you clicked,
everything was created with Claude. Now,
maybe you watched other videos and how you can edit videos with Claude. Cutting
silences and some motion graphic. But I
noticed that almost all these videos are missing the real steps and tweaks that make Claude actually replace a full video editor. What I will show you today
video editor. What I will show you today is a real engine I built that goes from a raw video till uploading to YouTube.
Everything is automated and you will not use any application, any video editor.
And yes, you are going to get this Claude editor for free by the end of this video. So, you can test it and see
this video. So, you can test it and see the real power of Claude video editing yourself. If you are ready, let's get
yourself. If you are ready, let's get started. So, let's go with a real
started. So, let's go with a real example. This is a raw video I want to
example. This is a raw video I want to edit. Step one is the cut. In this step,
edit. Step one is the cut. In this step, we remove silences, filler words, bad takes, so the video becomes ready for the actual work. So, here inside Claude,
I will simply tell it cut the video. And
while it's working, I want to mention something really very important about this step. Here, I'm using AssemblyAI to
this step. Here, I'm using AssemblyAI to transcribe the video or the audio. No
affiliate, no sponsorships, please focus. I saw many videos using Whisper.
focus. I saw many videos using Whisper.
Good, but with the new models from AssemblyAI, Claude will be able to understand your tone, your talking style, and cut with way more confidence.
So, in short, I transcribe with Assembly AI and feed it to Claude with the cut skill I created and it fully clean up my video and make it ready for step two.
And here I want to understand that every step coming next relies on step number one, the cut. If this step was wrong, everything else will not be accurate.
Okay, so that's step one. The video is ready. Now watch what will happen in the
ready. Now watch what will happen in the next steps that nobody shows you. Step
two is the visuals and here where everybody stops at the surface. Some
motion graphics like title cards, some text flying. Nice, but that's the easy
text flying. Nice, but that's the easy part. Watch this one. This is a VS Code
part. Watch this one. This is a VS Code window, not the real one. It's fully
generated with Claude. It's a full UI clone and it can simulate a realistic screen recording. So now I don't need to
screen recording. So now I don't need to screen record anything. Claude can
generate this for me. But here is my favorite one. Look at this. The cursor
favorite one. Look at this. The cursor
moves, it clicks, the page changes, the URL changes. I never recorded this. It's
URL changes. I never recorded this. It's
all generated with Claude with this skill I made here. So the only thing I actually record is my face. I hope you like it. Everything else on the screen
like it. Everything else on the screen was built, not captured. Now it's time for step three. Listen to this. This is
me raw and this is the same line after the clean audio skill runs voice isolation. Same me, but the voice is
isolation. Same me, but the voice is clean. Perfect. And this one matters
clean. Perfect. And this one matters more than people think. Nobody leaves
your video because a cut is bad. They
leave because your voice sounds like a bathroom. Now step four is sound, SFX,
bathroom. Now step four is sound, SFX, and this a game changer. For music it's somehow easy. You get one track under
somehow easy. You get one track under everything and you are done. Claude does
this too, but sound effects are hard because they have to land on the exact word. This is why the SFX skill reads my
word. This is why the SFX skill reads my narration and my visuals and plans every sound against a real timestamp. Let's
listen. That sound is one word, not near it, on it. And here's the part I like the most. Those sounds don't disappear.
the most. Those sounds don't disappear.
They go into a library named, sorted, saved. Same for the screen recordings,
saved. Same for the screen recordings, images, logos, everything I use will be saved, too. So, the next video, Claude
saved, too. So, the next video, Claude doesn't build everything again from scratch. It just picks them from my
scratch. It just picks them from my library. This way, the editor gets
library. This way, the editor gets faster the more I use it. Video 10 is way quicker and cheaper than video one, reusing the same assets. So, I built
this project, I built this tool, and now it's building itself. Now, step five is packaging, the title and the thumbnail.
And you already saw this one when you clicked on this video. What happens is Claude generates three thumbnails with three titles for AB testing. And again,
it learns and optimizes your next videos. Now, one thing I want to be
videos. Now, one thing I want to be honest with here, I'll be giving the project for free, but the models are not. Assembly AI for the transcript,
not. Assembly AI for the transcript, Eleven Labs for the music and sound effects, Nano Banana for the thumbnail.
They are cheap, but not zero. But a full video, fully edited, can cost you like maybe three, four dollars, whereas a video editor charges you 10 times this
amount for one minute. Sorry, video
editors, but that's reality and you should adapt. Now, step six, the upload.
should adapt. Now, step six, the upload.
Yes, also Claude, using YouTube API, will push my video. But before the upload, it checks itself. It looks for the words that got cut in half. Think
about it as a final review. It looks for some bad words, make sure that all you didn't drift away from my mouth, and then it uploads it directly to my
channel. From a raw file into YouTube.
channel. From a raw file into YouTube.
And since you reached this point, tell me please honestly, did you like the editing? If yes, hit the like button. If
editing? If yes, hit the like button. If
not, tell me what you didn't like in the comments, so I can improve in the next videos. The project is free, link below
videos. The project is free, link below in the description, and it's open source. So, I'm going to keep updating
source. So, I'm going to keep updating it. Transitions, effects, filters, and
it. Transitions, effects, filters, and much more. So, keep an eye on it, and
much more. So, keep an eye on it, and give it a star on GitHub. And if you want to learn how I build with AI, check out my full course in the description below. Thank you, and see you in the
below. Thank you, and see you in the upcoming videos.
Loading video analysis...