YT YouTube2Text

YouTube2Text → Transcripts → Claude just Edited an Entire Vox Documentary From Scratch! (Here’s How)

Claude just Edited an Entire Vox Documentary From Scratch! (Here’s How) — Transcript

by Joseph | Video Editing · 3,801 words · 556 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00Claude can now officially create Vox
0:02style documentaries like this one
0:04completely from scratch. Everything
0:05you're seeing right now, all six of
0:07these Vox style animations were created
0:10entirely using AI. And even if you've
0:12never used AI to edit a video before, in
0:14this video I'm going to show you
0:15everything you need to know to create
0:17Vox style documentaries like this using
0:19nothing but Claude. First, we're going
0:20to start from zero and create the story
0:22line and voice over using AI. Then I'll
0:24show you how we trained Claude to think
0:26like a real documentary editor and plan
0:28out animations. And finally, we'll use
0:30Claude and Hicksfield to create all six
0:32of these stunning documentary animations
0:34from scratch with sound design as well.
0:36And by the end of this video, you're
0:37going to be able to use Claude to create
0:39stunning documentaries like this
0:40completely from scratch, even if you
0:42don't know a single thing about editing.
0:44[music] And real quick, if you'd like to
0:45learn AI video editing from A to Z,
0:47we're building a full AI editing
0:49fundamentals course inside Ultimate
0:51Editors, but more on that later in the
0:52video. Now, before we create the crazy
0:54animations, we first need the story line
0:56and voice over that guide our
0:58documentary. So, let's quickly do that.
0:59Every documentary needs a good story.
1:01And typically, Vox covers political or
1:03historical problems backed up by
1:05numbers, facts, and cool stories. What
1:07makes their videos special is that these
1:09numbers and stories are brought to life
1:10by this rough paper style that makes it
1:12feel like you're watching an animated
1:14movie. Now, I know you clicked on this
1:16video to see how the vlog style is
1:17created with Claude, so [music] I'm
1:18going to keep this part quick. Here's
1:20how we created the story line inside of
1:21Claude. I asked it for a 60-second Vox
1:24documentary script. It covers one
1:25subject across six different settings.
1:27It has to include facts, numbers,
1:29graphs, historical event, and a simple
1:31story just like Vox. I let Cloud process
1:33and it gave me this. The script was
1:34really high quality and definitely felt
1:36like a Vox video. And on top of that, it
1:38included a real variety of scenes. A
1:40section with numbers, a section with a
1:42graph, a human example, and even a short
1:44story. Now that we had the perfect
1:45script, it was time to generate the
1:47voiceover. Vox documentaries are known
1:49for the cinematic voiceovers. It almost
1:50feels like a narrator telling you a
1:52story. To get that same experience, I
1:54took my script and headed over to 11
1:56Labs. 11 Labs has a lot of AI features,
1:58but the one that we need is this one
2:00right here, the text to speech. It turns
2:02any script into a highquality audio
2:04recording of someone reading it. So, I
2:06clicked on it, pasted in the script that
2:07Claude just made. Then on the right, I
2:09chose a voice that fit the documentary.
2:11I decided to go with Alex the business
2:13book narrator. Set the speed to medium
2:15and [music] clicked generate. After some
2:17processing, it created this amazing
2:18voiceover that actually sounded like a
2:20real documentary, which meant that now
2:22we had both the story and the voiceover.
2:24It was time to start creating the Vox
2:26documentary edit using AI. But before
2:28Claude can create a stunning Vox style
2:30edit, we first need to train it on the
2:32Vox style. You see, out of the box,
2:34Claude doesn't know how to edit a Vox
2:36documentary. But that's not a problem
2:38because we can teach it. We already have
2:40the script of our story. So, the first
2:41thing Claude needs to learn is how to
2:43look at a script and plan out
2:45documentary animations and scenes based
2:47on it. Here's what I did. First, I
2:49downloaded 10 Vox documentaries that I
2:51really liked. All of them full of
2:52animations that are built around [music]
2:54what's being said in the script. Then, I
2:56headed into 11 Labs and clicked on the
2:58speechtoext function. This transcribed
3:00any video and gives you its script. I
3:03clicked transcribe, select the YouTube
3:04tab, and pasted the URL of each video.
3:07After processing, we get this, the full
3:09script of the video with timestamps,
3:11exactly what's being said and [music]
3:13when it's being set. Then I click on
3:15export, export as a JSON, and download.
3:17We now have 10 documentaries with all
3:19their animations, and we have their
3:21exact scripts with what's said and when
3:23it's said. So, I fed all of this into
3:25Claude and told it to study these
3:27documentaries and their scripts, analyze
3:29what's being said, look at the video,
3:30and analyze what's being shown, and
3:32deduce why. Claude processed all 10
3:34documentaries and created what we
3:36[music] call the Vox editor brain. It's
3:37a complete guide on how the editors at
3:39Vox think when coming up with the scenes
3:41they want to show. And you're going to
3:43see just how important this skill is
3:45later in the video when we create the
3:46six stunning animations. Because it lets
3:48us feed Claude any script and it
3:50automatically plans every scene for our
3:52documentary exactly how Vox does it. The
3:54best part is if you also want to create
3:56a Vox documentary using AI, you don't
3:58actually have to train Claude like I
3:59did. You see, I asked it to turn
4:01everything into a skill that you guys
4:02can download and tell Claude to save.
4:04And every skill that we create in this
4:06video [music] is going to be available
4:07inside my AI editing fundamentals course
4:09in Ultimate Editors when it [music] goes
4:11live. If you're an editor that wants to
4:12learn how to edit all kinds of amazing
4:14videos with AI, check out Ultimate
4:15Editors. The link's going to be down
4:17below. So, now Claude can take a script
4:19and plan animations. The next step is
4:21turning that plan into Vox style
4:23designs, which means we now need to
4:24teach Claude how to design scenes just
4:26like Vox does. How on earth are we going
4:29to do that? Well, when you look at Vox
4:30scenes, they have a lot in common. They
4:32use a certain design language, common
4:34fonts, common colors, and common styles.
4:37We just need Claude to look at these and
4:38analyze the design system that Vox uses.
4:41Here's how. First, I needed to collect
4:43references of the Vox design. So, I head
4:45into Pinterest since I can find a lot of
4:46designs there and search the term Vox
4:48documentary animation design. I spend
4:50around 15 minutes collecting 100
4:52different highquality Vox designs that
4:54all fit the style. And I placed them all
4:56in this Figma file like so. So now we
4:58have a wall of Vox designs that Claude
5:00can study. Then I took the Figma PDF of
5:03this inspiration wall, uploaded it to
5:04Claude and prompted it to study every
5:07one of these designs and find the
5:08patterns and design systems that make up
5:10the Vox styles, the colors, shapes,
5:12text, objects, macro designs, micro
5:15details, and textures. And after letting
5:17it process, it came up with this
5:18in-depth PDF called the Vox design
5:20guidelines, which broke down exactly how
5:22Vox designs all of their scenes.
5:24Honestly, this was the most important
5:25part of creating the six stunning Vox
5:28animations, as you're about to see in a
5:29bit. And of course, I asked it to turn
5:31all of this information into a second
5:33skill for you guys, the Vox design
5:35brain. There's just one thing to keep in
5:36mind. Claude knows the Vox style, but
5:38Claude can't actually generate images.
5:40This is why we use a platform like
5:42Hicksfield, which can generate images,
5:44video, and audio. Since Claude knows the
5:46design and Hicksfield can generate the
5:48images, we just need to connect Claude
5:50to Hicksfield where Claude teaches it
5:52the design and Hicksfield executes the
5:54image. If you haven't connected
5:55Hicksfield to Claude yet, it's super
5:56simple. Go to this website, copy the MCP
5:59link, then head into Claude, click
6:00Customize Connectors, [music] add, add
6:02custom connector, and paste the link
6:04here. Now, just look at how far Claude
6:06has come. It can now look at a script
6:07and plan out every animation that needs
6:09to be on the screen and when using the
6:11Vox editor brain. And then it can take
6:13that land and turn it into highquality
6:15Vox designs that fit the branding using
6:17the Vox design brain. Just stay tuned to
6:19see how easy these skills make it to
6:21create the six amazing Vox style
6:23animations that I'm going to be making
6:24later in the video. But now this means
6:26that it's time for the final step,
6:28turning these designs into highquality
6:30Vox style animations. How on earth are
6:32we going to get Claude to do that? Take
6:34a look. Vox animations have a lot in
6:36common. The choppy frame rate, the
6:38movement, the transitions, the simple
6:39camera movement. This means that just
6:41like design, Vox uses a certain system
6:43for animating their scenes. But we can't
6:45get Claude to watch the animations. So,
6:47what I did was go through multiple
6:49animation scenes from Vox and manually
6:51explain to Claude everything that's
6:52happening. Every Vox scene breaks down
6:54into five major assets. There's the
6:56text, which usually has this minimal
6:57text pop-up animation. Then there's the
6:59main object, which typically has a paper
7:01unfolding animation or some kind of
7:02position pop-up. Then there's the
7:04background, which doesn't really
7:05animate. Then there's the secondary
7:06objects. These are just here to add more
7:09design and they usually have some kind
7:10of subtle static animation. And finally,
7:12there's the camera movement which
7:13sometimes has a zoom in or sometimes a
7:15camera pan around the scene. After
7:17explaining all five of these to Claude,
7:19it understood the assignment and built
7:21the third skill, the Vox animator brain.
7:23This was the vital skill needed to
7:25create the six stunning Vox animations
7:27at the end of the video. I'll show you
7:29how we built them in just a second. And
7:30just like the designs, Claude can't
7:32actually create animations, but
7:34Hicksfield can. So Claude is the brain
7:35that understands how Vox animations are
7:37created and it directs Hicksfield to
7:39create the animations themselves. So
7:40with all this training wrapped up,
7:42Claude can now do three huge things.
7:44First, it can look at any script and use
7:46the Vox editor brain to plan out the
7:48exact scenes we need to create for the
7:49viewer. Second, it can take this plan
7:51and use the Vox design brain to create
7:53Vox style designs that look amazing and
7:56fit the branding. And third, it can take
7:57those designs and use the Vox animator
7:59brain to create animations that look
8:01exactly like the ones Vox creates in
8:04their videos, which now officially meant
8:05that Claude could theoretically create a
8:08full Vox documentary if we give it the
8:10script. And that is exactly what we did.
8:12But before we put it to the test and
8:14create the six stunning Vox animations
8:16that you saw at the beginning of the
8:17video, you might either be an editor
8:19that wants to learn how to use AI in
8:21your workflow or even just a beginner
8:22editor that just got into video editing
8:24and feels overwhelmed with how much
8:25there is to learn. Let me tell you, this
8:27is completely normal and we all had to
8:29go through it. I recently interviewed
8:31our students and they all mentioned that
8:32learning editing on YouTube is
8:34overwhelming and unstructured. That's
8:36exactly why I decided to build Ultimate
8:38Editors 2.0. It's the best platform for
8:40beginner editors to master the viral
8:42editing styles of 2026 and make $2,000
8:44per month in 90 days. But now we added
8:47something even better. First, you're
8:48going to get access to the Premiere Pro
8:50and After Effects master classes to
8:52learn the editing programs step by step.
8:54Then you're going to get access to the
8:55viral editing style breakdowns of 2026.
8:58The cinematic style, the minimal
8:59animation style, the SAS motion design
9:01style, and the 30 plus viral animation
9:03breakdowns. Then on top of all of this,
9:04you're even going to get access to the
9:06$2,000 per month editor blueprint. So,
9:08you're not only going to learn the
9:09editing programs, master the viral
9:11editing styles of 2026, but you'll even
9:13learn how to make $2,000 a month as a
9:15video editor. But that's not it. With AI
9:17advancing fast, we're all struggling to
9:19keep up and learn how we can incorporate
9:21AI into our workflow to make us better
9:23editors. That's why I created the AI
9:25editing fundamentals course, where I'll
9:27teach you everything about editing
9:29amazing AI videos like this box
9:31documentary from scratch. And this
9:32course is launching in about 2 weeks.
9:34[music] If this sounds interesting and
9:35you'd love a step-by-step course on
9:37video editing, the viral editing styles
9:39and AI video editing, it's all available
9:42inside Ultimate Editors 2.0. The link is
9:44going to be down below. Click the link
9:46and we'll see you guys inside. But now
9:47that Claude was trained and ready, it
9:49was time to put it to the test and
9:51create the six stunning Vox animations
9:53that you saw at the beginning of the
9:55video. With Claude fully trained and all
9:57the skills inside, here's how we create
9:59a Vox documentary using nothing but
10:01Claude. Before I start, I first prompt
10:03Claude to load up all three Vox
10:05documentary skills. The editor brain,
10:06the design brain, and the animator
10:08brain. Then I tell it I'm going to ask
10:10it to edit a Vox documentary for me. It
10:12all starts with step one, planning out
10:14the edit. I upload my script and I
10:16prompt Claude to read through it and use
10:18the Vox editor brain to plan out every
10:20single scene for the video, just like a
10:21Vox documentary. After some processing,
10:23it thinks exactly like a Vox editor and
10:25creates a full plan of every scene it's
10:27going to create. Here's exactly how it
10:28works. Every phrase that's being said in
10:30the script gets a breakdown of what the
10:32big focus should be and based on this, a
10:34general guide on what the design needs
10:36to include. My documentary starts with a
10:38phrase about how in 2004 the average
10:40person could focus for 2 and 1/2
10:41minutes, whereas now it's only 47
10:43seconds. So Claude planned to introduce
10:45a year counter, then transition to a
10:46character on the left with text on the
10:48right highlighting it, and then to
10:49finish it off with a clock and text
10:51overlapping it. This was literally
10:53perfect. The second scene was about how
10:55this is happening from Tokyo to Toronto
10:56across the globe. So Claude planned a
10:58map animation of both cities, then
11:00transitions to a global map with red
11:02circles mapping out each city. The third
11:03scene was about attention span dropping
11:05yearbyear, exactly in 2007 when the
11:08iPhone dropped. So Claude planned a
11:10graph with a circle around the year
11:122007, then transitioning into a micro
11:14shot of the graph to emphasize this. My
11:16fourth scene introduced the average
11:18worker, how they check emails a lot, and
11:20how the reason is that the apps are
11:22designed to do that. So Claude planned
11:23an introductory scene revealing the
11:25character in the middle. Then two object
11:27scene showing the character and a
11:28MacBook and finally transitions into a
11:30big circle that closes the whole scene.
11:31My fifth scene talks about how the
11:33character's workday is divided into
11:35percentages. So Claude planned a circle
11:36graph around the character highlighting
11:38every percentage is needed. Finally, my
11:40sixth scene was a concluding scene with
11:41a quote. So Claude planned to simply
11:43write it on the screen with some extra
11:45design assets. This plan was more than
11:47perfect. Claude thought exactly how the
11:49Vox documentary editors do and created
11:51scenes based not only on what's being
11:53said in the script, but also on how we
11:55want the viewer to feel throughout the
11:57video. Now, with our plan created, it's
11:58time to bring it to life. This brings us
12:00to step two, creating the Vox style
12:02designs. I prompt Claude to use the Vox
12:04design brain to design every scene it
12:06planned exactly how Vox would, the rough
12:08paper, the textures, and the fonts. I
12:10tell it to use a blue teal accent color
12:12because I think it would look good and
12:14use Hicksfield to create the design
12:16mock-ups. It processes for a few minutes
12:17and then it starts sending me designs
12:19for every scene from the table. For my
12:21first scene, it gave me this design for
12:23the year counter, this design for the
12:24character and text, and this design for
12:26the clock and text. All of them matched
12:28the Vox style perfectly, which was
12:30exactly what we needed. For my second
12:31scene, it generated this map image of
12:33Tokyo, this map of Toronto, and this
12:35finishing global map scene. This could
12:36have been done much better, but it still
12:38did an amazing job. For my third scene,
12:40it created this vog style graph with a
12:42highlight around the year 2007. Then
12:44this micro shot of the graph with the
12:46year circled and it looked amazing. For
12:48my fourth scene, it created this
12:49character introduction with highlighted
12:50text, this sideby-side setup with the
12:52MacBook, and this concluding circle
12:53closing scene, and this was perfect. For
12:55my fifth scene, it came up with this
12:57simple character and piraphph diagonal
12:58design, and I honestly loved it. And
13:00finally, for the sixth scene, it created
13:02this text focused design, which was
13:04awesome. Using the Vox design brain that
13:06we taught it, Claude could create
13:08stunning designs that looked exactly
13:09like Vox. And honestly, this was just
13:11mind-blowing. Remember, if you guys also
13:13want access to these skills and want to
13:14learn how to edit videos like these
13:16using AI, the link to Ultimate Editors
13:18is going to be down below. You're
13:19definitely going to want to be inside by
13:20the time that we release our AI editing
13:22fundamentals course. Now that every
13:23scene has its design, it's finally time
13:25to turn them into full animations and
13:27create the Vox style documentary, which
13:29brings us to the final step. Step number
13:31three, creating the Vox style
13:32animations. Here, I prompted Claude to
13:34use the Vox animator brain to turn every
13:36single one of those scenes with its
13:38creative direction into a Vox style
13:40animation using Cance 2.0 in Hicksfield.
13:42I told it to make sure to use the rough,
13:44choppy, jittery Vox look on all the
13:46animations and for each scene to
13:48transition between the designs to make
13:50the edit flow a lot better. Now, this is
13:51the step that costs the most money in
13:53credits. But don't worry, this whole
13:55edit from zero cost me around $20 in
13:57Hicksfield credits, and that's with
13:59testing. Cloud first processed the three
14:01designs for scene one and created this
14:03amazing year pop-up character reveal and
14:05text highlight that transitions into the
14:07clock scene at the end. This looked like
14:08something straight out of a Fox
14:10documentary. For scene two, it created
14:12this map highlight that transitions to a
14:13different city and zooms out into the
14:15global map while highlighting the two
14:16big countries with red circles. It could
14:18still use some work, but it looked
14:20awesome. For scene three, it created
14:21this graph drawout scene with a line
14:24draw, then slowly transitions into the
14:26micro shot of the graph at 2007. The
14:28transition was rough, but the overall
14:30animation looked amazing. For scene
14:32four, it started with a paper crumble
14:34reveal of the character in the middle,
14:35then transitioned to the two object
14:37scene with the MacBook before zooming
14:39out and closing it all in one circle.
14:41Really solid flow. For scene five, it
14:42created the reveal on the left side with
14:44the circle graph, drawing out on the
14:46right side, which all looked great. And
14:48finally, for scene six, it finished it
14:49off by writing the text out,
14:51highlighting it, and having a hand icon
14:53pop in from the bottom of the screen.
14:55This was crazy. Using the Vox animator
14:57brain, Claude was able to nail the
14:59animation style down to a T. And with
15:01all six animations created, all I did
15:03was download them, import them into
15:04Premiere Pro with the voice over, align
15:06each one where it belonged, and add a
15:08mysterious synth wave soundtrack that
15:10felt like Vox. And with all this
15:11complete, this is what our final edit
15:13looks like. In 2004, the average person
15:16could focus on one thing for 2 and 1/2
15:19minutes. Today, 47 seconds.
15:24This isn't [music] happening in one
15:25place. From Tokyo to Toronto, every
15:28country that got smartphones got the
15:30same problem. And researchers can map
15:33the moment it arrived. Average [music]
15:35attention span, measured yearbyear,
15:38doesn't decline gently. It falls off a
15:41cliff. Right here, 2007. [music]
15:44The year the iPhone shipped. Meet the
15:47average worker. She checks emails 74
15:50times a [music] day, not because she
15:52wants to, because every app is
15:55engineered to pull her hand back. Here's
15:57her workday. 60% fragments shorter than
16:0110 minutes. Only 3%. This sliver is deep
16:06focus. So, the real question isn't
16:08[music] how we lost focus. It's who
16:11profits from never giving [music] it
16:12back. Ladies and gentlemen, creating
16:15this Vox edit taught me one thing about
16:17AI. It's a lot better when people
16:19actually know how to use it. If you're
16:20concerned about AI ending video editing
16:22jobs, I wouldn't be worried about that.
16:24And I'll make a separate video on why.
16:26What I would be worried about is other
16:27editors learning this before you,
16:29creating better edits than you, and
16:31ultimately closing higher quality
16:33clients than you. Because AI has done
16:35one thing to the editing space. It has
16:37killed the need to learn complex styles
16:39and programs. Instead, now you just need
16:42to understand content psychology, which
16:44is how good content is edited. You need
16:45to understand good design, which is what
16:47makes a good scene design and what
16:49doesn't. And ultimately, you need to
16:50understand how to edit videos with AI.
16:52That's exactly why I decided to create
16:54the A toZ AI editing fundamentals course
16:57in Ultimate Editors, where I'm going to
16:58be teaching you everything about AI
17:00video editing from the AI programs to
17:02the training to the skills. On top of
17:03this, our editing fundamentals course
17:05also teaches you how to design good
17:07scenes. And our Viral Talking Head
17:08Master Class teaches you advanced
17:10editing psychology. Thanks a lot for
17:11watching all the way to the end of the
17:12video. Stay tuned for our next banger.
17:14We'll see you guys soon.

About this transcript

This page contains the full transcript of Claude just Edited an Entire Vox Documentary From Scratch! (Here’s How) by Joseph | Video Editing, generated from the public captions YouTube serves with the video. The transcript has 3,801 words across 556 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.