Zapier and ChatGPT For Synthesia.io For AI Videos | Tutorial
Let's build with Zapier and AI (100+ videos)📅 2023-04-07
Transcript
[00:00] Hey, and welcome to today's video. We're going to be looking at Synthesia and using that API, integrating it with Zapier and ChatGPT in order to create automatic AI videos that are custom scripts and all work in the back end without you lifting a finger. Once you set up the flows, start off, go ahead and click the link below and you're going to be able to create a free Synthesia account, no credit card required. And it's going to take you to a nice little landing page. Get a better idea of what the soft where's and what we can do with it. Let's go ahead and jump over to that landing page and explore it a little. All right, as you see here, we come to the nice landing
[00:32] page. You can watch this little demo for two minutes. You can create a free AI video, no credit card needed. You're going to get access to a couple of API tokens. So you can go ahead and play around the software a little bit. But as you'll notice here, what's great about Synthesia AI is your ability to create at scale, but also create high quality videos that's going to be able to allow you to push out a lot more for your business or personal use on essentially of how to use this software. So we got training videos, how to videos, marketing videos, and so on. And what you'll notice is we have access to a ton of
[01:03] different avatars. And what you'll see in this tutorial is different ways you can utilize these avatars. But even further, you're going to be able to use this text and translate it automatically into these different languages and have it seamlessly work and be able to create custom templates that the API connects to and using Chat GPT to then feed it the script. Without further ado, go ahead and click that link below and create your free Synthesia account today. All right, as you always know, we are back into Zapier here. We're going to create our Synthesia API. If you want to know how to integrate Synthesia, AI and Zapier
[01:37] all together, go ahead and click that video right up there. Shows you how to implement AI automation and use it yourself. Zapier is a free tool. You get access to up to 20 Zaps for free. But let's go ahead and start building out this Zap. So for the trigger today, I'm going to set it up as basecamp, as we always know, to make it a manual trigger. In the context of using this as a social media scheduler, you would probably want to change this to Schedule by Zapier, choose a day, how often, so and so forth. But for the purpose of this video, we're going to go ahead and choose a manual trigger. We're going to do a new to do, and then from there, we're
[02:09] going to go ahead and choose our dummy account and we're going to be able to see essentially starting off and building out this zap. This is a great debugging tool, so you don't have to necessarily wait for a specific circumstance. You can manually execute the zap. Therefore, we can see the entire AI video be created live. Let's go ahead and do this as we go. So we created a trigger and essentially what we're going to want to do here is we're going to want to create a script. But before we do that, let's go ahead and jump over to Synthesia, learn a little bit about the software, create a
[02:39] template, and then jump back over here in Synthesia. As you see, you can go to your videos that will be currently live generated, your library and your avatars. And there is a ton of avatars to choose from. What I would suggest you to do is use the most updated and recent avatars. They're going to be called V three S. But for now, we're going to go ahead and create a template to be used in this video. So I'm going to do templates and I'm going to go ahead and say create new template. Let's go ahead and name it Synthesia Video.
[03:12] And from here you have a nice blank screen here. You can choose the templates that you want to work with. So there are pre made templates. Let me go ahead and show that real quick. You go to feature templates. You can actually choose from all these different templates, essentially the type of video you want to create. You got banking templates, office interior. You can go to click this and it gives you a nice little tutorial. You can go to click the video to see what it looks like. And as you see here, it starts with some nice clip art.
[03:44] Kind of skip through this. And for the sake of this video, we'll go ahead and use this template so we can get a better idea. So I'm going to go ahead and go ahead and say this little point right here. And typically what you want to do is click it, say use templates. And with the $30 plan, you get access to up to ten scenes. In this context, we're going to create a three scene video. So let's go ahead and jump in some templates to use here. So we'll go ahead and go to templates Synthesia.
[04:18] And this is going to be basically templates you've already made, you can kind of grab from. And as you see here, we can actually jump around. So maybe you don't like all the cards. You can grab cards from different areas, but for now, we're going to use the banking. We're going to go ahead and use this scene and let's go ahead and find our closing scene, which is going to be right here. All right, perfect. So to start, we are going to create three different panels here, showcasing the three different possibilities here. So we're going to make this for Joe's Coffee.
[04:48] You already know we love joe's coffee. So we're going to go ahead and change this image to that logo. Go ahead and upload that. All right, we are uploading the logo and we can go ahead let me go ahead and add the title here and video template YouTube. And we're going to go ahead and change this. So essentially, like we've seen before with other API tools, the way it works is that your fixed variables come like that and then obviously you can add variables on the
[05:21] screen and add variables in the actual underlying script. Let's go ahead and change this. Real quick is I'm going to add text. I'm going to add a title here. We're going to do joe's coffee. Go ahead and make this a little bit smaller here. Let's go ahead and do 90 perfect. Go ahead and make this white. And let's spell coffee correct. And then this. So this is going to be a fixed variable here. For our sake, we can go ahead and unlock this. So we can edit this.
[05:52] And let's go ahead and real quickly create a description for Joe's Coffee. Just thought I might as well show you a quick way to create a description so we can go ahead and do business. Name, not chat. GPT Joe's Coffee Service. We are a cafe that sells coffee. And then we go ahead, generate two sentence description.
[06:24] Enter. A lot of times when you're dealing with GBG four, your underlying input doesn't have to be too complex relative to what 3.5 we had to do. So in this context, it kind of knew what we wanted to do here. So we're going to take this line, bring it back to Synthesia. And this is just for the beginning. Obviously, for our flow, it's going to be all automatic. So we're never going to have to go back and forth like that. Again. For this context, we'll go ahead and do that. And we're going to go to 25 here and let's go ahead and take this and make this the variable. So we're going to go ahead and unlock this.
[06:55] We're going to delete this. And for now, you are going to become my variable. So I'm going to unlock this and we're going to do title. So to make variables in Synthesia and essentially just do little brackets like that. And now we have our title. You're going to want to lock that so that it understands not to push over to this side. We have some media being used here which you can replace like this. And what's great about Synthesia is your ability to essentially put live video as long as just regular video.
[07:26] So maybe we want to use this video, people are happy, but or we could actually switch that up and let's do unsplush coffee and let's do cafe, see if what we got going on here. This doesn't look bad. Double click to change the formatting. You want to zoom in. Push this to the right a little bit. Okay, perfect.
[07:56] And then you can obviously change the color by just clicking, doing the fill. Use this little sticker. Grab that. And then maybe I want to do this as well. Let's go ahead and grab the same color here. And then let's go ahead and slightly change it like that. Okay, so we have our initial screen. What we want to do here is basically this is fixed text. So fixed text will be like every single time. This will be set every time it's generated. But let's make this completely variable.
[08:26] So we're going to do semi or bracket. Bracket. We're going to do introduction bracket, bracket. And that makes it a variable. Perfect. And from here you can choose your avatar. So to do that right now, it's Voice. So we're going to just choose Alex. And for the initial screen, we're going to just choose Voice only. And then the voice we'll use is professional. So let's move to the next screen. All right, so from here we have a nice little quote, a background, so on and so forth. What we want to do here is we can go ahead and change this to the colors used in the past.
[08:59] So let's go ahead and grab that real quick. Grab that. Put that here. If you feel like you've learned anything, make sure to like this video. It really helps us here at Web Cafe subscribe, as we do daily videos like this, showing you how to use GPT and automation and really scale your business and overall lessen your workload. So from here, we're going to go ahead and change the color of these two. And then you can make this fixed so we can just say
[09:34] quote of the day. And then what we can do here is we can just go ahead and change this to just the variable text quote, and go ahead and lock that. So we maintain within the parameter. And then we're going to add another script here. So we're going to do quote script. And then we'll do our final scene here. But before we do our final scene here, we're going to go ahead and grab that avatar that we just had. Let's show a different use case. So we're going to go ahead and use her as a little circle
[10:07] here. Delete this, because that is not necessary. We use this as a little circle here. And then maybe you want to change her background to these backgrounds. We can copy that. Grab this paste there. Okay, cool. So we have that and then we have our final little screen here. Let me go ahead and change the color to our branding. Like that. Okay, perfect. And then just change your different needs.
[10:37] So what I can do here to make it faster for myself, so I can go ahead and go like this. Shift command c come back over here, command V. Just make sure both selected. Perfect. Thank you for watching. And maybe we want to make this all just fix variables e. And we can do our address here. Perfect. And then for our final conclusion, let's go ahead
[11:09] and add some animation here. So we're going to go bracket, bracket, conclusion, bracket, bracket. And then you can add pauses here, get markers for the animations. But for today's video, we're just going to do one thing. We're going to add a little gesture here, which is going to be ahead. Yes. And then as you see here, we have our full made template. Now one other thing you can do here is when we can add music to it, we can add transitions. So if you wanted music, we're not endless happiness for the
[11:40] transitions. We can do anything provided here. So let's go ahead and do swipe down for this one. Let's do slide up and then end it like that. And then what's really great here is you can go ahead and hit generate. This is going to create the template. And we're going to go ahead and include captions. And from there we're going to convert the video to template. And from there, let's go ahead and see how we take all this information, bring over to Zapier, use AI and
[12:11] generate a fully AI video. All right, let's go ahead and start. So to start off with any great idea is you're going to have to get the concept behind the idea. So we're going to do ChatGPT here. We hit conversation. And we're going to just do rename this as video idea. Hit continue, continue. And let's go ahead and make this a specifically tailored video here. So we're going to do a couple of things. First thing we're going to do is to say you are a coffee
[12:42] expert. Let's go ahead and fine tune our scope. Then we're going to add a memory key here. We're going to do Coffee vid idea. Then up here, as we know from previous videos, you can want to choose a more complex or more comprehensive model, which is going to be GPT Four. And then for the message, let's go ahead and create essentially what we want going on here. So we're going to go ahead and say generate a video idea. Or let's do this, generate a video idea for a brand named
[13:15] Coffee or Joe's Coffee and have it focus on a daily coffee fact. Let's go ahead and see what it comes up with. And as you know, with this kind of stuff, always testing, always debugging. So maybe this first execution isn't what we're exactly looking for, but we can work with it. So let's go ahead and see what we get here and generated. And as you see here, look at this.
[13:47] Too much of what we're looking for here, right? We didn't necessarily want the entire shabam. So let's go ahead and add some constraints here. Go ahead. Parameters. Generate one coffee fact. Max one sentence. Sometimes you have to wire in. And what I like to do is, when the output is way off base, just add a character here. So we start a new chat. So in the context of ChatGPT, it's like starting a new chat, clean slate. So it doesn't have the preexisting chat that we just made
[14:20] there to kind of make it more biased on what its new output would be. So let's go ahead and see what this will be. You see, here we have it. So we got title joe's Coffee, daily Fact video idea. Discover the amazing benefits of Joe's Coffee. Did you know? Okay, for the sake of this video, in theory, I could do something really cool here, but we kind of get into more complex ways to parse data in our social media management course. But in theory, I could parse this data so it separates the title and the video idea, therefore saving you on
[14:50] expenditure. But for the purpose of this video, since I want to give a more general basis, we're going to go ahead and try a regeneration of this with a slight tweak on the underlying prompt. Hit two. We're going to say, generate video for a brand named Cop Joe's Coffee and have it focus on a coffee fact. Generate okay, max one sentence. Generate just the coffee fact. No text before or after. Sometimes you have to get really precise here. But what's great about GPT, since it is a human language
[15:22] model, is it can really understand and comprehend what you're trying to say when you get a little bit more detailed. So let's go ahead and see what this says. All right? The lighter the roast of the coffee beans, the higher caffeine content. Perfect. Just the coffee fact. That is what we're looking for here. So from here, we got a couple of things to build out. First thing we got to build out is going to be a title for the video. Maybe you can use this title for YouTube, Twitter, so on and so forth. So let's go ahead and create a title for this video. We do ChatGPT, Be, and for our sake, actually, let's make
[15:53] our lives easier. We can go ahead and duplicate. And why are we duplicating? The reason we're duplicating is because we've already set some parameters here. We've set that you are a coffee expert. We've set the model, so on, so forth. Let's go over and rename this to Video Title. It's going to be used for the social media, and it's going to be used within the actual underlying video. So I'm there. We're going to go ahead and delete that now, video title, you could use 3.5. You could use four for stuff that I feel is more imperative to the content. I like using four just because and I'd rather use something
[16:28] that's going to take a little longer to have better content because in the day you could be anywhere. Now you could be at Chili's ordering some ribs. Because automations, that's what's great about automations, it's all automatic. And now with AI, we're making this stuff even better. So we're going to say based on this coffee fact, parentheses assistant response message, generate a enter your social media. For this context we'll say YouTube, YouTube video title.
[17:01] Let's go ahead and see what that looks like. And we're going to go ahead and change our key here as we do not want the same key here because it just had the previous prompt. We don't want to add another completely different prompt in the same chat log as it'll throw off outputs and make them less accurate. Here's our YouTube title discover the Power of Light Roast. Unleashing the Maximum Caffeine your content or in your coffee. And as we know, and as you've seen in my previous videos, we got those nasty little quotation marks. But with one little block we can solve that. So we're going to do formatter by Zapier rename this remove
[17:33] quotation marks the content or sorry, the event we're going to be doing is manipulating text. We're going to grab the text that we just generated here. We're going to hit replace because we obviously want to replace the quotation marks with nothing. So we hit show all options and we're going to go ahead and find the quotation marks. Continue test action. And what you'll be seeing here is the exact same content both at the quotation marks, which makes for easy doing.
[18:04] All right from here. So we have just created the video title. We have the initial inception of the video which is going to be based off the fact now we need to create an introduction that is going to be the script used in the video. So let's go ahead and I'm going to delete this. I'm going to go ahead and duplicate this again and we're going to use outline here and we're going to just go ahead and drag this around if it allows me to, there we go. Boom, boom. And then we're going to rename this to introduction.
[18:39] And from there we're going to create a little introduction here. So we're going to say based off this coffee fact, generate a YouTube video introduction. Now if I just leave it at that, you're going to probably get more than you bid off, more than you could choose. It's going to generate probably too much text than what I'm personally looking for. So in this context, I'm just going to say parameters generate a max of three sentences. And one thing you'll notice with GBC is really pushes those sentences to the limit.
[19:10] It'll use its grammical abilities to really make an elongated sentence that's still grammatically correct, which is helpful and not helpful in some situations. So sometimes making it do a max of two sentences really like layman wise, looks like four sentences. But let's go ahead and continue here. We're going to retest this action. Let's go ahead and generate ourselves the introduction to this YouTube video that is specific to the topic at hand, which has to do with the coffee fact of the day, which is that the lighter the roast of the coffee beans, the higher
[19:41] the content. So we got our hey there, coffee lovers. Now, in theory, you don't have to worry about these quotation marks if you're not using captions in the video. I could you add a formatter block just to remove those? I could keep testing this to not. But in a day for just vocal, you actually won't see it. You would only see it if you keep the caption on, as you saw in the Synthesia video. So for this case, we're not going to go ahead and continue that. We're going to go ahead not remove the quotation marks because it's not necessarily pertinent, because the Synthesia doesn't say quotation mark. Hey, no, it just goes.
[20:12] So we're going to go ahead and add another block here. And this is going to be the quote description blocked. And in this context, quote will be the fact. So we'll just say fact description, and this is going to be the dialogue for the fact description and for the underlying block of text that was a variable. We'll just put the fact and this will all make sense once you see the full generation. For now, we'll just do this fact description. Okay, perfect. And we're going to say based off this, go over here, click.
[20:45] You'll notice is you're basically building blocks here, right? Based off this YouTube introduction to a coffee fact, generate a YouTube video. What does it say? Based off based on this introduction to a coffee fact, generate the next part describing
[21:15] the use cases for the fact. Now, you can obviously manipulate this a little bit more, but I'm actually real quickly going to edit this a little bit further here. Sometimes you got to do this. So instead of that, we're going to actually have our baseline be the coffee fact. So understands what we're referring to here. So I'm going to grab the coffee fact again. And then what we're going to do here is introduction to the video dialog. And then we're going to grab the introduction. And now what am I doing here? I am essentially giving GPT
[21:48] context of what's going on here. And then essentially now it's context. Now it knows what the coffee fact is. Now it knows the introduction to the video, the dialogue that was just said. So now it can actually say the next part of dialogue and sound seamless like it isn't just two separate executions. So generate the next part describing what the use cases describing. Generate the next part describing use cases for this fact.
[22:18] Let's go ahead and see what GPT comes up with. We're going to go ahead and rename this to Coffee Fact as we want to start a new specific dialogue conversation on a new chat, new clean slate, new execution, new output, let's go ahead and see what this comes up with. All right, so as you see here, it came up with in this video, we'll explore why those early morning pickup me ups. Now this is good. Problem is that I don't like this in this video because in the day we're halfway through the video, we don't need an
[22:50] oversight, right? So we're going to say generate the next part. Describe the use cases. Fact, just describe use cases. No introduction. See if that helps it tailor itself. Obviously, change the chat that we were just in. So we're going to add a one. Let's go ahead and see what the new output would be. All right, so let's go ahead and wrap this with a bow and duplicate this. And what's so nice about Duplication is it speeds up the process once you build out your overarching context that you're dealing with here. Conclusion, and this is going to be the last little speech
[23:22] here. Now, we could reference more information such as the company and stuff like this, but for the sake of this video, we're going to do this. Just the final little speech here for the slide introduction to video dialogue. Okay, perfect. We're going to do description. Do your description of the dot of the fact. Same deal here. Let's give it context what was just described. And then all we want to do here is we're going to go ahead
[23:54] and say generate the conclusion for this video. Let's see if that's enough context for it to understand what I'm referring to when I say that. Change the title of that real quick. I'm going to retest this and let's see what the final script would look like. Perfect. In conclusion, not only do coffee roast impacts flavor profiles, it keeps going on. So maybe I want to add one little interesting thing here. Maybe I want to say ask the watcher or say viewer to subscribe. And like, maybe we want to make this a little
[24:27] bit more tailored out of a CTA, but we don't want the CTA to essentially be a generic static text. So every time it says it's a little niche specific variable text and then you add the static like and subscribe. No, we don't want that. We want it to ask in its own way to make it as natural as possible to help you out. So we're going to say inclusion lighter roast packs a more powerful caffeine punch, making them excellent choice for these moments when you need an energy boost. If you enjoyed this Coffee Insight, please don't forget to hit the subscribe button and give you a video a thumbs up.
[25:00] Until next time, coffee lovers stay caffeinated. That's really good. I really like that. Very specific, very niche. Make sure you like it. If you feel like you've learned something in this video. So from here, let's go ahead and tie and build this present with our nice final little block. Obviously you can push this further and go a little bit more crazier, but let's go ahead and generate this Synthesia video. So we're going to shoot new video from template. Continue, continue. And from here we're going to see some nice variables. We need inputs. So let's start off with the template ID.
[25:30] All right, to get the template ID, all you got to do is come over here, hit copy ID and paste. And then what I'm going to do real quick is essentially all those little variables that I created. You got to enter them in here, make sure they're spelled correctly. Capitalization is important, but I'm going to go ahead and do that real quick for this video, all right? So I went ahead and grabbed all those different variables we created here as seen here. Don't add the curly brackets. Capitalization and uncompromisation is important. So make sure you do it exactly how you put it into the video. So let's begin here. So let's go ahead and add our title video title.
[26:02] Real quickly, let's add the formatted version of that title so we don't have the quotation marks, the introduction, grab our introduction here, assistant response message, our quote and do the video idea. Just going to be quote in this context is the fact being dealt with our quote script. So we're going to do the description system manager and the conclusion. And then we're going to conclusion here. And then from here all you got to do is hit you can
[26:33] obviously maybe make the video title a variable so when it shows up in your Synthesia or your description maximum out of words here, maybe use three to five. So in theory you can use a GPT 3.5 summarizer tool here, summarize that title so you can have an idea of what's being produced here. We get more into that in our social media management course. From here you can add a callback ID. So callback IDs are great. So when you're executing a lot of different templates on a lot of different topics, you can add filters to other Zaps. So let's say I want to use this video in YouTube, post it
[27:04] automatically and post it as a short. So I want to do a couple of things here. I want to make sure that we're calling onto the right flow so it knows essentially this is metadata. So I could say this is coffee fact. Then I would create a separate Zapier that would filter for Coffee Fact Synthesia videos that are ready. And then from there I would go ahead and automatically crop it to a 916 ratio made for TikTok and YouTube shorts. And then from there I would automatically publish to YouTube with a title, description, specific niche,
[27:35] hashtags, so on and so forth. We jump into. All this crazy stuff into our social media management course. So make sure to check down the description. But for the sake of this video, let's go ahead and hit continue here and let's test this action. So we went ahead and sent the video over. Right now it's rendering out the video. This takes a little bit of time. What I'm going to do in the meantime is I'm going to go ahead and make sure that this video is put in the description so you can look at it for reference. Also, the Zap app in today's video is going to be in the description and then other resources that you could possibly use such as the Synthesia, such as our course on specifically this kind of stuff for social media management, and furthermore, different things that you
[28:07] could find of value, such as different ChatGPT prompts. All right, so as I said before, you can find this video in the description below. Let's go ahead and click it and you'll be able to hear the audio if you watch it below. But as you see here, we have our title here. Now obviously it got feedbacked over, but that would just change a little bit of the underlying sizing of the box and locking it different ways so you can just manipulate the underlying template on that. But as you see here, we have the nice description below,
[28:37] then we have the quote. Now in theory, what I would probably want to do is move this to the left a little bit more. What you'll notice with this type of stuff is I'm going to go ahead and mute this real quick. What you'll notice this type of stuff is you're going to want to edit and change the template as you go. There's going to be a lot of trial and error when it comes to this type of stuff. But as we see here, we have this our nice little conclusion speech is happening right now. Our static variables are occurring here and then just goes over everything you want to know with the music. So I'm going to go ahead and share this, check it out in
[29:08] the description so you can hear the audio. This is just a general idea of using AI automation ChatGPT and connecting it all to make video content. You know how to use AI automation to make an AI video specifically for videos, and different ways you can execute different scripts, how they layer on top of each other, how ChatGPT communicates within the Zapier. If you want to learn how to do this with images and make an AI automated tweet with an AI automated image that goes along with it, really cool stuff. You can find out in this Banner Bear video right there.
[29:38] But without further ado, go ahead and check out other videos on this channel. Go ahead and subscribe, as we do daily content here, when it comes to AI automation using ChatGPT and other softwares. But without further ado, I'll see you in the next video.